Welcome to this one-off evidence session on the future of warfare. I am very pleased that we have an eminent and distinguished panel in front of us today. I am very grateful to Dr Keith Dear, CEO and founder of Cassi, for taking the time to give evidence today. It is wonderful to welcome back Air Marshal (Retired) Edward Stringer; thank you for your private evidence session with us earlier, and we look forward to your contributions as a senior fellow at Policy Exchange. It is also wonderful to have with us Sir Hew Strachan, who is Wardlaw professor of international relations at the University of St Andrews. Let me kick-start proceedings with a question for you, Air Marshal, about the wider strategic context. There has been a lot happening in terms of warfare, but can you set out briefly the key trends and relationships shaping the future of warfare? Could you also talk about the future of an international legal framework?

Thank you to the Committee for giving us all the opportunity to give evidence this morning. To answer your question, I will canter very quickly through three background factors, which I think we have all, in one way or another, been writing about for at least the last 10 years, but some of which have only become salient in the last couple of years. The geostrategic backdrop—the geopolitics—is the convergence of three events. The first is climate change, which we have never dealt with before. The second is the rise of China and the shift in economic power to the east. With economic power comes political power and the ability to invest in military power. That is probably a once-in-500-year event, given that 500 years ago we saw the rise of western Europe and the centricity of that region. The third is the merging now of what some call the fourth and fifth industrial revolutions. These things are all interrelated. The fourth industrial revolution is a move from the industrial age to the information age—the digital age—and the fifth is, largely as a result of climate change, a move to a sustainable energy economy and therefore moving beyond the era where all economies were hydrocarbon-centric. Of course, those two emerged in the high-tech era we live in today. What does that mean for the military? Military power always follows the industrial revolution. A hundred and fifty years ago we had the first industrial-age war, the American civil war, since when we have had wars of industrial production and wars of the railroad, the factory and the telegraph. All that revolutionised the way that armies fought and nations went to war. Now that we are through the fourth and fifth industrial revolutions, we are going to see a change in the way that armies fight and nations go to war. I could simplify that by picking on four technologies, without going into detail.

Some commentators have highlighted that we might have a return to state-on-state conflict. Do you feel it is more nuanced than that, in terms of the use of proxies or things happening in a grey zone sub-threshold?

It is a bit of all of those. Against that backdrop, it is the way that states manipulate things that matters and shapes how we must respond as a nation state and with our alliances. Very briefly, on how the fourth and fifth industrial revolutions are coming together, you now have enhanced connectivity. We are all used to it: the battlefield cloud allows data and associated machine learning to be present at the edge in our warfighting capacities, and it allows artificial intelligence to do things that were previously science fiction, which allows autonomous systems to come to the fore. All that is reshaping what we termed, nearly 10 years ago, the economics of lethality. In the industrial age we built big industrial platforms such as submarines and fighter aircraft—all the technology was in the platform. Now you have smart warheads, swarming warheads and swarming weapons, which cost a lot less than the platforms. That is the economics of lethality. To reference what you have just said, yes, we are back from the slight holiday from history that began in 1989. With the rise of China, suddenly the west is facing a competitor that it has not seen for 500 years, and this audience knows exactly what I mean by that.

We will get into China later. My request to all the panellists is to be as concise as possible because there is a lot of ground to cover and we want to get views from all three of you. On most of the areas we want a short, concise contribution from each of you so that we can move on to the next topic. Hew, what do you think are the key trends and relationships shaping the future of warfare? As an eminent academic, you have studied many of these things, so would you also speak to us about the future of an international legal framework with regard to those trends?

On the first point, I would simply refine what Ed said. This point comes out of the Changing Character of War programme that I set up and ran at Oxford for more than 10 years: you need to recognise that old trends continue alongside new. It is not all new, and the tendency is always to be led by the new technology. That does not mean that the new technology is not important—it is, but some of the changes can be changed back, rather than taken forward, and some of the changes have to accommodate existing platforms. No armed forces start with a tabula rasa of totally new equipment reflecting the state of the art. To cut to your point about the international legal order, this is hardly news given what has happened in the last 12 months, but it seems to me that the post-1945 international legal order has collapsed, to all intents and purposes. It has not been helped by our own participation in something called the rules-based international order, which seems to me a profoundly unhelpful label. It locks people into an idea of a US-dominated world, it does not appeal to the global south, and it does not make sense of what is going on. Within law, we have a crisis in the relationship between the decision to go to war and how you actually fight a war. In the history of international law, since the 18th century, those two things have been treated separately. Now we see a thoroughly justified Ukraine fighting a war of national self-defence against an aggressor. We have accumulated evidence of Russian war crimes—about 150,000 cases, at least, as recently as a year ago, stored in the cloud—as far as Ukraine is concerned. There is no hope of dealing with that number of cases in a legal system. The consequence is that the question of Russian responsibility for war crimes at individual soldier level essentially gets folded into the argument that the war is unjust in the first place and Russia should not be fighting it. We have a flipped argument in Gaza, where Israel is seen as thoroughly justified in its response to Hamas. Hamas, of course, is not a state actor, so from that point of view it is outside the UN charter framework, but Hamas clearly launched an attack that would be seen in UN terms as a breach of the charter if it were a state actor. However, there has then been a disproportionate response in terms of the conduct of war. That matters enormously for the United Kingdom because we have positioned ourselves as a middle-ranked power that is ready to defend and depend on the international legal structure that we have created. That is what gives us influence. We need to address how we can put that back into shape, but at the moment, we are not addressing it. I realise that you want to get on, but I would make two points very quickly here. The first is how we handle the distinction in international law between the justification to go to war and how you fight a war legally in terms of its conduct. Those are now in tension with each other, although the issue has always been paradoxical. The second issue is that we need to enable our armed forces to have political utility. In other words, they need to be able to fight in ways that are legally defensible. That reintroduces a concept that has been essentially written out of international law, but was very powerful in the 19th and early 20th centuries, which is the idea of military necessity. It is not just that you need to be justified in fighting the war in the first place, but that you need to enable your armed forces to do so in ways that are effective, if that serves the ultimate objective, which in our understanding would be a defensive struggle.

That was a fascinating contribution. The reason I allowed you all to set the scene is that I think it is important to get those things on the record before we delve into each item. Dr Dear, given the wider strategic context and my question about the key trends and relationships, what are your thoughts on an international legal framework, and how do you view future warfare in terms of the use of proxies or things happening in the grey zone? Please set the scene with respect to your own academic work on this.

I think the first and central theme that drives a lot of the urgency in geopolitical competition is the fear of the arrival of artificial general intelligence. There is a tyranny of definitions here but, broadly put, I mean an AI that outperforms most or perhaps all humans in all cognitive domains. Nature has this morning claimed that AGI has been achieved. That is quite a bold statement, but it has gone so far as to say that it has. Everything we care about in defence, as well as much more widely, is the product of intelligence. Weapons are the product of intelligence. Tactics and strategies are the product of intelligence. If you have access to a smarter model that can give you better tactics, strategies, weapons and so forth, you are, by definition, going to have a significant advantage over your adversary. That is true of competition in the military domain, but it is also true of competition in the economic domain, and it is true of Governments as well, with much more efficient and effective policies. With that as a backdrop, you see a fierce competition in the US with a very different model—backing the largest AI companies to build the largest possible models, with as much money channelled into those models as possible—and with a view that perhaps there is a winner-takes-all outcome here, where if you have that smarter model, you nationalise it. The US presidential memo for AI under President Biden has a clause for the US to nationalise any AI company that gets to AGI, which I think is super-interesting. In China, the fear is what it would mean if the US got there first and the Chinese were locked out, so their strategy is an AI-plus strategy. They are not able to invest at the same rate, and they are not innovating at the same rate; consequently, the majority of their investments are in fast followership. That does not look like a bad bet. The open-source models they build are six to 12 months behind the models released from the US, so they are consistently behind, but they are consistently following. Were the US to hit a threshold where they do not go any higher—were there to be a top of the S curve, rather than continuing straight lines on graphs—China could win. Again, we then have a nation that is more effective than we are in building weapons and is coming up with better tactics, strategies, policies and so forth. I have only really talked about China and the US because, to be honest, in terms of winning the race, they are the two major players. If we take a UK perspective, we have an AI action plan and we have some things that are beginning to address this, but in defence, we have a huge say-do gap—to use Edward’s phrase—between the profundity with which we talk about the importance of AI and the reality of AI deployed in UK defence. I am sure that we will come back to that. For me, the thing that is driving the maniacal competition in energy, for rare earth minerals, and more broadly for economic dominance globally, is this fear of what artificial intelligence will mean militarily, economically and politically for all of us.

That is absolutely fascinating. I want to ask a second question to get the views of all three of you, because I want to get things on the record, but if you could be fairly concise, that would be helpful. You mentioned our closest ally, the US, and we have all seen that the US Administration have changed course with the new national security strategy and the defence strategy. What do you think are the implications of that? In your esteemed opinion, do you think that UK defence should be planning for a future with less co-operation with the US, or will that not be the case?

I think there is much more ambiguity in the national security strategy and in the national defence strategy of the US than we—the press, in particular—have read into them. For me, from the UK point of view, the crucial points are, first, that we are not separately mentioned; we are part of Europe. Furthermore, the Americans see us as, and we must become—to use the vocabulary of the defence strategy—partners not dependants. So the drive is there, and if we react to that challenge, the implication is that we will be seen as good boys, or girls, in the eyes of the United States. In other words, it is not a closed door, as I read it. The other point I would make is the issue to do with extended deterrence within that. There is no statement about or reference to extended deterrence in the defence strategy—in fact, the whole defence strategy is extraordinarily thin on actual capabilities as opposed to statements of intent. The strong stress is on deterrence by denial, which is exactly what got us into problems in relation to Ukraine—a reluctance to use deterrence through the threat of escalation, which might actually have deterred Russia, or at least met Russia on its own terms. Those seem to me to be the principal takeaways for the UK. What I conclude from that is that, as far as the United States is concerned, the Indo-Pacific is not an area of interest or a concern for European powers. There is no reference to that. There is a lot of reference to allies in the region. I do not think there is actually a reference to Australia, despite AUKUS, but there is obviously reference to South Korea, Japan and the immediate neighbourhood. With this Administration, to be dogmatic about the future is clearly difficult, but I take that as the document is written. What I think is interesting about both documents—it is really striking—is that alongside all the praise of Trump and his single-handed driving of all this, which of course you can dispense with, is the clarity of expression and the directness. To imagine the UK, let alone a previous US Administration, writing documents that speak so simply and clearly, and reference old concepts in ways that Americans can understand—the Monroe doctrine being the obvious one—is, for us, extraordinary. We would struggle to write a strategy that was that clear and that simple, yet that of course is precisely why it will have power.

It definitely makes very sobering reading. Dr Dear, very briefly, what are your views about the implications of the US national security strategy and the defence strategy?

I think they are a continuation, in one case explicitly, of the Hamiltonian and mercantilist view of the world in terms of competing continuously for relative power with your allies as much as with your adversaries. You see an element of McKinley in the talk around Greenland—back to the future, as it were, in terms of the acquisition of territory and dominance over those territories. I think the UK is not prepared to compete in that world. We should look really closely at what the US says in both its national security strategy and its national defence strategy, then ask a deeper question about why it says it. Specifically, the Americans talk about how the Indo-Pacific region will be half of the global economy. There is a reason why they would like us to be less active in the Indo-Pacific, which is that there is a relationship between your military co-operation and your ability to secure the kind of economic collaboration and co-operation that you need. The world’s largest middle class is going to be in the Indo-Pacific. If you move to a much more automated economy, yes, you can now produce goods independent of or at least with reduced—vastly reduced—dependence on demography but, in the end, robots do not buy goods and they do not pay taxes, so where are you going to sell those things? It is going to be where the economy is growing fastest and the population is growing fastest, and that is the Indo-Pacific. We should be really careful before we assume that some of the things we are being asked to do in terms of burden-sharing are purely about fairness, as opposed to being about dominance and relative power in the international system.

That is very thought-provoking indeed. Briefly, Air Marshal, what are your thoughts? Will we be able to rely on our US ally going forward, or are we looking at less co-operation?

We have to buy a seat at the table. The national security strategy has rowed back its description of China: China is now just another competitor in science and technology—the economic battle that Keith mentioned. In fact, as part of the writing team for the integrated review, he referenced it a couple of years ago. If we want to be taken seriously, we have to compete in that battle. That means we have to understand the risk we are taking here. You ask whether things change, and we have seen this before. In 1945, remember, we were thrown out of the nuclear programme by our ally, America, after we had put all the MAUD Committee’s great intellectual property into it. Attlee and Bevin decided, “Right, we’re going to have a bomb and we’re going to have a bloody Union Jack on top of it.” We need to compete, we need to fight, we need to take risks and we need to invest in all these technologies, at which point America will see us as a player. You could argue that they are now having a proxy battle. China is supporting Russia to fight Ukraine in Europe. I was in Berlin just before Christmas and this has really shocked the Germans. In the 28-point peace plan, suddenly the security architecture of Europe is being decided, and Europeans are not invited into the room. We have to fight, and we have to compete. America is still a democracy and a natural ally, but we have to approach it from a position of strength and be able to negotiate in a hard-headed way.

Thank you all very much for your contributions. I have a very simple framework question. What has come through overwhelmingly from what you have said is how remarkably unserious we are when facing the challenges in front of us and the nature of the escalating defence and security context. Without being too crass about it, could you score the UK’s seriousness between one and 10? Just give me some sense of that: are we a two or three, or are we a seven or eight? Where are you each coming from on our level of seriousness in responding to this challenge? The say-do gap that Ed has talked about strikes a very deep chord, so I want to know whether it strikes a chord with you guys.

Well, I would put the UK at the bottom end. The strategic defence review takes most of this pretty seriously, but where has it landed?

But in terms of actually doing things—

This is, of course, like any academic hedging, a response to a stray question, but if you look at when we could possibly deliver seriously, it would not be until after 2030, or even 2035. That is assuming we do everything we have committed ourselves to doing. What could we do in 2026 or 2027? Almost nothing. There is an enormous immediate gap, quite apart from the long term. Most of the responses about seriousness are long term and are felt in only certain sectors of Government. This is a national question, not a sector-of-Government question.

Great. Dr Dear, maybe you could answer next, as I think we probably know where Ed is coming from.

I think you have to look at this in different ways, but I will put a number on it. If you took the strategic defence review—although some of those things would need developing—and explicitly said, “How likely is it that we will have implemented the things that are listed in there by, say, 2030?”, I would say it is about a three out of 10 on your framework: about a 30% probability that we will have implemented those things in a way in which a reasonable observer would accept that we had. You can see that in the way that MoD talks about private capital, for example. You do not have to persuade private capital to invest if they believe there is a return there. You can see from that real, live, actual market, which more accurately represents bets on the future, what different people around defence think of the likelihood of the UK investing. To be slightly fairer to UK defence and the UK more broadly, if the question was about how seriously they are taking the threat within the resources they have and what they can do today, as opposed to the honest arguments on how likely they are to achieve for the future, that score would probably go up. I think there is a recognition that we face an urgent threat, but my probability on their actually doing something about it and the likelihood of achieving those things is much lower.

It is interesting that we both plumped for three out of 10. I would say that you have to split the UK up a bit, because we do have some extremely good AI companies; when you talk to the industry, it is massively frustrated. The problem is that we have a big, reluctant machine in Whitehall that is still trying to force this new world into the old processes and hopes to carry on muddling through. Given what we have all said about risk and the speed of change, you need a big transformation—dare I say it, you need a Haldane 1906 moment—and that has to be driven politically, with real political clout behind it. I do not think that any more money thrown into defence will necessarily be spent wisely until you make that change. You have to get that change done, and it has to be driven with a real sense of political clout.

So you are definitely in the three out of 10 area.

Yes, around there, given how the nation state and the machinery of government collectively are approaching it.

I am going to ask about potential collaboration between China and Russia, how close they might get in the future and whether we are ever likely to see a combined force fielded. What do you think has already happened in Ukraine in that space, whether it is tech or personnel? The context is that Russia is modern warfare experienced in a way that other countries are not, and China is churning out military capabilities at an astonishing rate, be they space or maritime capabilities, and personnel-wise it is in a different position from most other countries.

I have just come back from Ukraine, and they will point out that the Chinese have just discreetly built literal supply lines—railway lines—to drone factories in Russia. Open sources suggest that 60% of the Russian effort is being covertly bankrolled by China. Even though there is no ideological alliance—

Sorry, just say that last stat again.

About 60% of what is keeping the Russian war effort and economy going comes from China. There is a huge amount of support from China; it is just not in weaponry with a Chinese flag on it. There is a huge amount of support to Russia, and Russia can only maintain this war because China is essentially bankrolling it. That is because it is in China’s interests to keep the west and America distracted. Russia is useful to China at the moment. If you look at weapons tech, there has been an inversion there as well. The previous generation of Chinese aircraft and ships were copies of the Russian ones. That is not the case now. The stuff that is now being developed—the latest generation of Chinese warships and aircraft—is their own manufacture. Obviously there has been a lot of intellectual property theft from the west, but the boot is on the other foot, and at the moment, Russia is very useful to China as a distraction that is keeping the west tied up.

In 2018 I wrote for the RUSI Journal in response to Putin’s statement that whoever leads in AI will rule the world. That statement was hyperbolic at the time, but it increasingly looks like that is going to be true. I said then that Russia is weak and weakening in technology. Its technical talent is leaving, and its educational system—once the pride of the world—is also weakening. The chances of it leading in those kinds of emerging technologies are only diminishing—and diminishing faster since the start of the Ukraine conflict. What that means is a growing dependency on China. It is absolutely useful for China to keep the west occupied in Ukraine. It is also quite useful for China to keep Russia occupied in Ukraine, because it is increasing Russia’s tech and economic dependency, and therefore increasing China’s influence over Russia, reducing the threat that China might face from the north, and enabling it to increasingly leverage Russia should it need to. It puts Russia in a position of greater dependency. Therefore, I think alliances are likely, but they are not going to be anything like equal partnerships. That is in part what drives the US’s view that maybe it can prise Russia away from China in the medium to long term. Whether that is right or not is a separate question that I am happy to tackle, but what you are seeing is ever-growing dependency, and the question is how far Russia is willing to accept that growing economic and technological dependency.

That is very interesting. The picture from that answer and the previous ones is that because Europe, including the UK, is falling so far behind on AI and other tech advancements, and on spending power, it feels in a way—although you have not said this exactly—that the Ukraine-Russia war is becoming a proxy war between the USA and a China-backed Russia. You guys have not said that, but it is an interesting theme.

I think “proxy war” becomes an unfortunate label, because you do not get into the nuance of what is going on. To row back, the logical argument has always been that Russia and China will fall apart at some point, and you need to exploit that. The trouble is that China at the moment does not seem to want to have a conversation with Washington. At least under the previous Administration, efforts to use the moment created by the war in Ukraine to have that conversation were not being responded to from the Chinese end. Whether Trump can be more successful in that domain still seems very much an open question. From the point of view of the war in Ukraine, I do not see any desire on China’s part for the war to escalate. It was very clear at the beginning, when it responded to Russian nuclear threats, to say, “That’s not the line you’re going down.” In other words, that reinforces the argument that it suits China to have this war ongoing, and Russian dependency adds to that. The other point I would make about this relationship—this is a very Mackinderite geopolitical point—is that you now have a bloc of Russia and China, with Iran in the mix too, that is geographically contiguous, with extraordinary leverage from that position in what Mackinder would have called Eurasia, with a capacity—Ed’s point about railway lines makes the point—to communicate easily enough if the investment is made. That extends into the Stans as well, potentially. How you disaggregate those states, as opposed to encouraging them to work together, is probably not within our gift, in terms of our own power. It is in the power, possibly, of the United States. There is a point in the US national defence strategy where it talks, in a sign of potential leverage, about—I used Mackinder deliberately—the US and its allies being on the periphery of this bloc, but all round it and therefore having a degree at least of power over trade and power over what happens.

If there is a ceasefire in Ukraine in the next year or two, do we think that China would be encouraging Russia in some way to take further invasive action somewhere else on the NATO border, be it Finland or Estonia? Would that be in China’s interest?

My sense is that China is cannier than that, and that it would see where the opportunities lie. Ukraine was an opportunity. China did not instigate Russia’s attack on Ukraine, but what it has done is enable its continuation over a protracted period—

I know, but if there was a ceasefire, would China be encouraging Russia?

I think that the answer is, if that continues to result in cheap Russian oil coming into China, it would possibly have an incentive to do this.

Can I have very quick answers to that question from the other two witnesses?

I do not think it would want to foment conflict, but it would be very happy to see Europe tied up, spending a lot of its GDP on defence against something that looks like a frozen conflict around Ukraine, Belarus and so on.

I think the same. It is unlikely that it would actively advocate an invasion, but the idea that it would want Russia to be just as concerned about the European threat and therefore spending and focusing its war economy west rather than south and east, I think would be true. That is not the same as saying that I think it would be actively trying to precipitate an invasion or a direct hot war.

Let us move on to defence reform. Air Marshal, you have written quite a powerful piece about the say-do gap, which focuses on defence reform and some of its initiatives. We also had a report from this Committee, “It is broke—and it’s time to fix it”, which speaks about defence procurement and the “political imperative” that is necessary to translate our ambition and limited resource into some impact. We heard that defence reform would be the biggest obstacle to a successful SDR delivery. Which initiatives from defence reform show any sign or promise of addressing the say-do gap or the challenges that are put to us in the “It is broke—and it’s time to fix it” report?

I will rehearse very briefly the two I mentioned. The whole point of defence reform was to make named individuals accountable for actual outcomes, rather than there just being, over the road there, a bunfight over managing resource and therefore inputs. Given that we are talking now about state-on-state conflict—serious, prolonged war—in any war your reserve forces are vital, and it is interesting that the reserve forces in the UK report that they are suddenly being taken seriously, because now they are run by the operational output side of the Ministry of Defence and not just seen as an HR or personnel question to be run in the area of Chief of Defence People. Ditto, military strategy and ops is no longer just managing the current operations around the globe through PJHQ, but thinking about the sort of questions that we have discussed this morning. However, the big fiefdoms of the old model are still there: the big fiefdom of finance and military capability, the military strategy and ops, and the Chief of Defence People’s area. It has not yet turned into a properly military strategic headquarters that is looking at the actual outputs and therefore taking the three services properly under command. I think you are seeing that at the moment in the inter-service horse trading over what the defence investment plan looks like, which seems to me like the chiefs behaving just as they have before, without a military strategic headquarters dictating what vital outputs and actual outcomes are to be achieved. Without that, despite the reform of having a National Armaments Director, the National Armaments Director and his staff cannot start making the really hard-nosed decisions that would be necessary to get us through and demonstrate, not least to the Treasury, that the MoD is reforming its procurement practices. We talked before about whether we are taking this seriously. I think this is an example where we still have too much business as usual. Without a change, it will be hard for the Secretary of State to make the case to the Treasury that more money will be properly spent and will be delivered in actual outcomes. I think the Treasury would have a point if they said this will probably be soaked up and evaporate in the normal way. That is the problem with defence reform at the moment: it is stuck in a halfway house, and it has not got to the stage yet where you have those four named individuals sat around the Defence Board table in front of the Secretary of State, actually responsible for hard outcomes. What is the military posture to do, and is it achieving those outcomes in the strategic context that you mentioned earlier on? National Armaments Director, are we spending money wisely in delivering only the resource that we need rather than that which the services want? We have not moved that far yet.

Alex Baker will come in on some of the methods by which we deliver that output in terms of financing, so we will come back to you, but last week we were in Barrow, and we heard about how one of the major programmes was being asked for options, which is kind of evidence of that bad behaviour. We also put it to the CDS that the fact that the three individual service chiefs had gone out and made their own pitches for the world was absolutely perverse, and kind of laughs in the face of defence reform. I was going to go to Dr Dear, but, Sir Hew, you are nodding violently.

I am nodding in agreement without a significant contribution to make, I am afraid.

There are other examples where there is progress, but not enough, which builds on Edward’s point in terms of defence reform. An example is that the SDR talks about a novel tech ringfencing—moving 10% of defence budgets into novel technologies every year. We absolutely need that, and it could drive the kind of change we need, but because, as I understand it, there was some resistance to having a definition of what was meant by novel tech, there is a real danger that a lot of things that were already in the pipeline just get relabelled as novel and then, “Oh, look, we’re doing it.” When I left, I thought that the MoD system could not be reformed from the inside. I have not changed that view. There are other examples of things that superficially are fantastic, and then you think about them for a bit longer and they are not quite there. The defence industrial strategy has an accountability table. That is great—we have never had anything like that before, to my knowledge. It then lists people by position, rather than by individual name and what they are responsible for. The problem with that is that the tenure for most SCS is shorter than any of the programmes. Even with the new rules that say you are trying to get from six years to two years for contract award—not for new capabilities—most senior civil servants will be gone. Maybe 50% of people have changed roles after two years. Maybe there are only 30% left after three years. There is no alignment of incentives and individual accountability, and no one is actually named. I hesitate to say that it is a Potemkin effort at accountability, but it is not quite what it says on the tin. I think that that is a real concern, especially given the necessity of speed in an era when everything is accelerating, which I hope we will come back to at some point.

Can I just come in with a quick thought? Earlier, Ed referred to a Haldane 1906 moment, and the historian in me bristled a bit—as it does when Ed makes historical points. I simply want to say that what Haldane was responding to was the need to cut Army expenditure in the first instance and to prepare it for imperial defence. It was not a great plan to deal with the first world war. The same applied to the Navy. That makes a point in relation to Keith’s argument that pressure from the outside led to Haldane coming in with a bunch of fresh ideas, which he put through not as fully as he wanted, and without a clear vision of what was going to happen, but, crucially, to think about how it could be more efficient in the first place. I think that there are elements here where you are trying to construct what you are fighting for, which of course you want to do, and I am not saying that you should not do that, as you clearly need to plan. But there are things to do with process and outcomes that stand independently of any scenario building, because the scenario will almost certainly be wrong in its particulars. It might not be wrong in its general indication, but it could even be wrong in that respect—you might prepare for one sort of war and it is another sort of war. If you do not think about how to deliver outcomes quickly, you are certainly not in a position to be able to respond when the war does come.

There are three of us in the room who believe that the system could not be changed from the inside, and that is why the three of us are in this room. I would like to hear more evidence. I wonder whether Luke Pollard’s response to Ajax, and finally seeing an individual held accountable, is maybe a start in that shift—I see Air Marshal Ed chuckling. Are there perhaps some more examples? All I am seeing at the moment are old behaviours repeating themselves and coming out in many different ways. Prior to coming on to the money, one of my big concerns is that those behaviours do real, lasting damage—and Michelle can explain how that is, or was, the case in Barrow. Are there any other examples that we should be focusing on or looking at?

Are we looking for positive or negative examples?

Negative—of where we are seeing that defence reform is not doing what it says. We have had some good examples from Air Marshal Ed at the beginning. Are there other things that we should be looking at or focusing on?

I think we should have a laser focus on whether innovation is pulling things across the valley of death. We have never been short of low-cost pilots in multiple new technologies across defence—things that look good on the OJAR and sound good in press releases. But it is hard to see that on the frontline in the UK, and a meaningful change in how we fight and what we buy to enable how we fight. In the end, it is in that outcome. Is the frontline changing at the speed at which technology is moving? The answer to that is clearly no. I think that many inside the system would be willing to say exactly the same, so I would keep a laser focus on that.

Do you see any evidence, Air Marshal, of a willingness from the people to change? I have given an example, which I think I stole from some of your writing, of those three service chiefs coming out with individual plans.

Without getting into where we are currently with the Ajax programme, because it is obviously terribly murky and commercially sensitive and all of that, how did we get there in the first place? And, even now, how is the question of Ajax being framed? I would argue that it is framed as a procurement problem. Go back a couple of years, and other services and various others would always have a chuckle and say, “Oh, that is the Army’s problem to solve.” But it wasn’t. A military strategic headquarters, were it to exist, could not deploy the land component and meet our doctrine and our NATO commitments. Yet that was not the first and foremost conclusion around here and in the MoD. Instead, it was, “Oh, we have a procurement problem to solve.” Some people would argue with me over this, but you could run this again, going back 10 or 15 years to when the programme started, and essentially ask: has the Chief of the Defence Staff signed a land component doctrine that says we need a vehicle that looks something like what is now being satisfied by Ajax? “Yes? Right, now go to the market and see what you can get.” “The CV90 looks pretty good. It’s made with a huge amount of British input through BAE Systems. It works, and there are a load of allies”—that is the important thing, so the production line behind it will stay. “That’s perfectly good enough.” If we had done that, that would have shifted the Army not to an organisation that is trying to perfect kit—and becoming, as all the services are, a pretty amateur procurement agency. It would have shifted to being a fighting service: “This is what you get given. Make the most of it”—which is actually what the services are really good at. Our personnel are really good at making the most of what they are given. Defence reform will shift those priorities. Military strategic headquarters says what the priorities are. We can probably all list those now, and it is probably less contentious than it has been for a while, following what has happened with the United States recently. We could list those priorities, say, “These are the military options that need to be done,” then you hand that kit list over to the NAD and the stuff is given to the services. That inverts the way things work at the moment.

It will not surprise you to find out that I want to ask about defence finance. Our adversaries are redesigning their financial infrastructure to fight their war, yet our banks struggle to lend to defence in the fullest sense that they could. Can you talk about some of the innovative solutions to finance that the UK should be looking at, including the multilateral Defence, Security and Resilience Bank? Maybe you could comment on the news from Friday that Canada is anchoring conversations about the funding of that bank?

The Chair will be pleased to know that I will be very brief on this, because my colleague Dr Dear has hands-on experience. The first thing to say is that we still live in the era of ESG; therefore, it is hard to invest in defence. That needs to shift. That could be politically driven from the top. We need to get hold of money, and we need to get hold of money at very good long-term rates. Something like the Defence, Security and Resilience Bank is a very good idea for doing that. Our allies, including most of the JEF nations, are really keen on this. For some reason, it has been London—Westminster—and our officialdom that has pushed back slightly. You will have to ask them why. I am told that there is a bit of “not invented here”, but there is also a bit of, “That’s a funny finance model. We just go to the Treasury, Treasury gives us money, we go, ‘Time and materiel,’ and we buy stuff.” It is interesting that Canada has leapt in and said, “If London”—a global finance centre—“and the British Government aren’t going in, we see an opportunity.” This is an ex-central banker: if you want a vote of confidence in it, it should be the fact that Mark Carney is behind it. This is my final point. We talked earlier, Chair, about how quickly things are moving and our risk appetite. If we want to start being able to compete and say to our ally, America, “Actually, we’re just as big and strong as you are,” then this is a small amount of risk to take to provide us with the funding mechanisms to invest in that science and tech that underpins military power. If we cannot take even this amount of risk in going for things like the Defence, Security and Resilience Bank, what sort of strategic shock is going to make us move to a slightly different model? Can I hand over to Keith?

Just to add to that, the Defence, Security and Resilience Bank team estimate that, at the moment, our banks can lend £12 for every £1. They estimate that, with the credit guarantees that the Defence, Security and Resilience Bank could deliver, our banks could lend £30 for every £1. Do you think there is an understanding of these defence finance issues within the Treasury and the MoD?

I think there is a misunderstanding of what drives the capital gap. Principally, the view is, “If only we had more money that we could pump-prime, suddenly we would have all these technologies.” The problem is that nobody believes that the things that are getting built are going to be bought. There is not a problem with global liquidity or capital more broadly—maybe at series B in the UK, because of the broader depth of our market, but not in the early stages for companies like mine. The problem is convincing investors. That was true when I was at a large corporate with a major defence account, arguing to the global board for investment in technologies that we wanted to sell into UK Government and defence. It is true now as a start-up, having those conversations with global investors. They do not believe that the money is going to be there to deliver a return, and so far that negativity over the last five years has been proved right. There is not massive investment in technology; there are not contracts coming out. And when contracts do come out of defence, they come out in penny-packeted ways. I think the Defence, Security and Resilience Bank is something that we need. I think this debate is one that we need. But I think there is a misunderstanding of what is meant by “demand signals”, which seems to be the term du jour in the MoD. Demand signals are the MoD buying things at a scale that justifies private investors putting capital in. The problem is not that we cannot get the money to invest. It is that, to convince an investor, you have to show how other markets can be accessed, because nobody really buys the thing. I think that is a market judgment on the likelihood of our getting to 2.7%, 3% and ultimately 5% of GDP and that translating into investments in those kinds of emerging technologies at the scale that we all think we need. It makes me slightly sceptical of things like the £300 million going to NSSIF for defence. The National Security Strategic Investment Fund invests in niche technologies where there is no market—absolutely—but there should be a market in the Ministry of Defence, as a first customer, for some of these defence products. If there were, there would be no problem crowding in the capital. So yes, it’s a good thing. Is there a problem with a lack of understanding? To some extent. But I think the bigger problem is seeing this as a supply-side problem and not a demand-side problem.

Could I briefly answer your question about whether this is known about? I think it was in 2019, in the defence strategy group, that the idea was put across the table, because in some of the work I was doing then in dealing with dual-use companies, they would point out they would go to J.P. Morgan and talk about big data; because they now had a contract with J.P. Morgan, they could go to the markets. The figure I used was not your 30 to one. I just said five to one; you can get five to one, and the dual-use companies are telling us we should be doing this. And the response was sort of, “Well, that’s all very interesting. Move on.” There was no spark; there was no, “Oh, that’s interesting,” and I still don’t think they have grabbed it.

I would like to ask Sir Hew first on this one. Do you think that the UK’s posture on deterrence needs to change in the light of shifting ways of warfare?

Well, I think we need to begin by talking about it, which we don’t do. It is not in the public consciousness. For deterrence to operate, there has to be public ownership, particularly if you’re talking at the nuclear level, because you are actually involving the whole of society in the implications of the use of nuclear weapons, but ever since the cold war we have not wanted to have that debate. Given the fact that that is the biggest procurement item currently in the defence budget, it is extraordinary that we are not having that discussion and that there is not even public awareness that it is the biggest item in the defence budget. They are focusing much more on the stuff that gets the headlines, which of course is AI, AGI and so on. So I think that discussion has to begin, because there is a social dimension to deterrence; society has to be resilient. Bizarrely, on the recommendation of a colleague in what is now the Department of War, as opposed to the Department of Defence, I am reading a book called “Alas, Babylon”, which is set against the context of the 1959 missile gap and the Sputnik crisis, and there is a moment in that novel where the victims, essentially, of a nuclear attack on Florida—it is on the US generally, but Florida of course is less badly hit because it did not have significant military targets at that date; it had no Central Command headquarters—say, “If only the Government had spoken to us!” But they were so nervous about creating panic that they never had the conversation. I think we need to treat the public with more openness about that. The next point—I have made it already, in a different context—is that our thinking about deterrence is deterrence by denial because we are thinking in defensive terms. The trouble with that is that essentially the possession of nuclear weapons—I am still talking about nuclear weapons in this context of deterrence—is about an offensive capability. We simply cannot have a conversation that is limited in that way if we are going to extract value from this investment that we are making. The point about deterrence is that there is a capacity to escalate the conflict. So we need to have that discussion too, and to recognise that deterrence by denial when you have an adversary that, in the case of Russia, does use the threat of escalation—that has to be countered by a threat of escalation. The final point I would make in this discussion is simply this: where are we with our allies? Again, there needs to be a more grown-up discussion. Self-evidently, the understanding of a nuclear deterrent is that it is a sovereign capability, but, of course, in the UK’s eyes, we allocate it to NATO—in other words, it is a very significant part of our contribution to NATO. That has resulted in the last year in both Germany and Poland having to have conversations around this issue. It has in the past already, of course, produced a degree of Anglo-French congruence, which seems important. We then have to think, if nuclear deterrence is to have a much higher salience within this discussion, what sort of capabilities do we need? Is it simply SSBNs, which is the route we are still committed to? We made a virtue of the F-35As and their acquisition, but they were in the acquisition programme anyway. Do we really need a tactical or theatre capability? The French are very nervous about that, because they think it would lower the nuclear threshold in Europe—that is understandable; that precisely goes back to old debates about deterrence. We are improving defence in order to increase deterrence, but one of the challenges we have is that if something is unused, it gradually loses its deterrent effect—and, obviously, we want nuclear weapons to be unused. I am minded to think that this is a conventional parallel; before the 1982 Falklands war, the argument was that Britain would do very little and NATO would also do very little. In the end, of course, Britain did respond, and successfully, to the Argentine invasion. The response to that successful campaign against the Argentines was to say, “This has enhanced deterrence. It has put NATO in a much stronger place, because it shows that a NATO power is ready to use force.”

I am from Barrow-in-Furness, so we understand the deterrent very well. Do you think that the rest of society understands the importance?

No. I probably cannot refer to another Committee, but when I was an adviser to the Joint Committee on the National Security Strategy, we used to come back to this argument quite a bit. Many members of the Committee felt that, “Well, you don’t want to scare the horses, so we shouldn’t put it too high up the agenda, and if we do, anti-nuclear lobbying will become greater.” If you look at the MoD’s figures on public opinion—I have not seen the latest ones—there is a very wide understanding, even in areas of the United Kingdom where there would not be, including Scotland. The responses are not that different: most of the British public understand why there is a nuclear deterrent, and that is without having a conversation with them. We should be less frightened of that conversation, because if we are frightened, we do not prepare the public for understanding it. Of course, people will be worried about an all-out nuclear exchange—they jolly well should be; that is precisely why it has effect—but we need to be much more open in our debate.

What impact will the failure of support for the international non-proliferation agreements have on this, such as the expiration of the New START agreement?

Probably all of us at this table would want to have a constructive debate about nuclear arms limitation. Under the previous Government, this country was going to increase the number of warheads that we had, without fully explaining why we might want to do that, and without putting it in an arms-control narrative. If we do not take a lead on it, the argument for nuclear proliferation will gain a head of steam. We need to be seen to be active. But that does not so much require a conversation between Moscow and Washington, because, at least under the previous United States Administration, even when there were escalatory threats—I gather the UK and France were both included in this—there was a back-channel conversation between Moscow, London, Washington and Paris. Crucially, where is Beijing in that argument? That seems to me to be the big question in terms of nuclear weapons and their control.

We have seen an escalation in attacks in the grey zone over recent years; we talk about cyber-attacks and, again, I would suggest that we do not talk about that enough publicly. People might see a delay to their shopping from Marks and Spencer but not understand the implications of those attacks on our supply chain. I have also heard recent discussions about offshore oil and gas seeing more drones in their airspace. How should the UK develop its deterrence in the grey zone, and is the establishment of a new cyber command a good start? I direct this to Dr Dear.

The establishment of the cyber command is a good thing. My view is that we ought to have an independent cyber force and an independent space force. Until you truly have independence, you will never get the arguments that you need to resource proportionately to the threat you face. Those arguments have to run through single-service, career-incentive systems, where arguing that we should invest more in cyber or in space, at the expense of the thing your service does, is not beneficial for promotion. There is a fantastic paper called “Victory through Space Power”, which is just the old “Victory through Air Power”—the case made for the creation of an independent air force—but with the word “space”. You read it and think, “Well, if there’s a case for an independent air force, there’s an equally strong case for an independent space force and an independent cyber force.” I think it is axiomatic; the logic is irrefutable. I think it is only resisted because people are worried about losing some of their slice of the pie in the wider intraservice competition that it would generate, and because of broader change fatigue. The cyber and specialist operations command is very much a good thing. Personally, I do not think it goes far enough, and I suspect that some of the reviewers would agree with me.

And on developing deterrence itself?

On developing deterrence, it seems to me that we have a really perverse compromise in terms of political pragmatism, which allows us a logical deterrence blindness. Logically, if Russia is running a war economy now, and if China is saying that it wants to be ready to seize Taiwan, if it needed to, by 2027, it does not make sense to say, “We think we’ll be ready by about 2030, when we’re going to start investing enough.” Everything in deterrence theory would tell you that you are making the risk of something happening greater, because your adversary judges, “Well, we better act now before you’ve spent all that money and got there.” If you believe the MoD’s and the UK’s line on when we are going to spend this money, it would require you to say, “Hang on a minute; we need to acknowledge that, in what we are saying, we are accepting that we are increasing the short-term risk, because we are providing an incentive for others to act now.” That is true in cyber and in all domains. I come back to the maniacal need for more speed in everything we do. The speed of technological change is such that we saw our first fully automated cyber-attacks, run on Anthropic’s models, reported just before Christmas. We are likely to see ever more agentic cyber-attacks over the next year, run by AI only, at a scale we are in no way ready for. We need to move at the speed the US is in deploying models; we are talking about two years to get to contract for major defence programmes, but the US defence AI strategy, published on 9 January, talks about 30 days to deployment of the latest AI models, and about designing new capabilities within hours as requirements flow back from the frontline. That is hours in physical capabilities, but it is going to be minutes or seconds in cyber, where you just have to respond and build. There is a real worry about bending to the political pragmatism of, “We couldn’t possibly spend this much now, so we’ll be ready later. That should suit us, and we can balance our investment over time.” We need to respond to a threat that everybody can see is here now and is accelerating rapidly in technological terms. We create an incentive for others to act sooner if they know we are going to be ready later.

Does the public understand enough about what is happening in the grey zone, about deterrence and about what we need to spend on defence for us to justify that politically?

Very briefly, please, Air Marshal.

I do not think so. It was Michael Quinlan who spoke about the offensive spirit; that was probably the last time we had in the Ministry of Defence a civil service official and an intellectual who really thought through deterrence. My point, Chair, is that we actually need an intellectual rearmament in order to think through all this. That then needs to be transmitted to the general population. In closing on this point, perhaps I can link to one you made earlier, Chair. To hammer home what Keith has just pointed out, AI is very good at writing code. Things will happen really, really quickly. Our legal codes sit there and say, “There will always be a man in the loop.” This is one of those areas where we need to really think things through, and do it from a moral and ethical standpoint, in the way that Michael Quinlan looked at deterrence with nuclear weapons. It is easy to be defensive, and to be backward leaning—it makes us feel good—but if you lose that offensive spirit that Michael Quinlan used to write about, you have not got the ability to underpin deterrence and make people think twice. These are very complicated questions, and we need that intellectual rearmament so that we can gather them all together and think through the implications of them combined.

When we hit the point where AI can produce AI.

We are there now.

That’s what I’m saying.

Air Marshal, I want to talk about the lessons learned from Ukraine. The SDR says we have demonstrated that “a nation’s Armed Forces are only as strong as the industry, innovators, and investors that stand behind them. And that technological innovation is vital to stay ahead of our adversaries.” What priority lessons should we learn from Ukraine? Is that just limited to harnessing drones and data and digital warfare?

It is very much about the power industry. It is very much about the railways. The first thing I would say is that the lessons of Ukraine have to be evaluated by an independent commission; it cannot be left to the MoD. I won’t rehearse what Keith said, but people in the MoD are cherry-picking some of the answers because they support some of the cherished programmes. You need to put in place an independent commission to learn the lessons of Ukraine and to feed in, in the way that our Committee of Imperial Defence did between the wars, looking at these sorts of lessons and working out what they mean for mobilising society. That would then have to feed into implementation of what is recommended in the SDR, with some form of Bill tabled to support national mobilisation, national resilience and so on. We have got out of the habit of that since 1989, when we drifted out of producing the war book nationally and took it as a savings measure, which no one really noticed. There is another area there of intellectual rearmament. It starts with a broad-based, non-military commission that looks at the lessons of Ukraine and then feeds them back in, as we work out what it means to be a nation state at war. That is what underpins deterrence.

Can I ask you a specific question? How are drones affecting warfare today, as they have never been conventionally used before?

In my opening remarks, I talked about the economics of lethality; that is the point, and I will link that to where I closed my last point, on intellectual rearmament. I am staggered that I still hear officers in many places—not least when we declared IOC for Ajax, which was reported widely—when asked about drones and the battlefield we see in Ukraine, say, “Oh, but we wouldn’t fight like the Ukrainians.” That strikes me as a very complacent and rather arrogant statement. The future of warfare does not look exactly like Ukraine, but it certainly involves a lot of what you are seeing in Ukraine, adapted for future theatres. Just look at what Israel did in managing to achieve air dominance over Iran for 12 days. It did that with some of the conventional capabilities we have in the UK, and with some of those clever drone-based, cyber-based and intelligence-led capabilities that you see in Ukraine. At the moment, I do not see that blend of thinking applied to the way we do force development for our own forces.

To build on Air Marshal Edward’s point, we have seen a really interesting narrative arc in relation to Ukraine: it went from a starting point where people were arguing, quite loudly, that drones did not really matter—they were just a side issue and were not really making any difference—to people suddenly saying that drones are really important, but only as an augmented capability to large crewed systems, and now we need autonomous collaborative platforms that sit alongside and support. That fundamentally misunderstands the nature and rate of progress again. We have gone from systems that were doubling in agentic capabilities every seven months, to agentic capabilities doubling every four months. What that means in real terms is that, in 2023, AI models could complete tasks that took humans about 30 seconds—to 50% of what a human would do, but the point is the trajectory—while in 2025 they could complete tasks that would take humans an hour. On the current trajectory, by 2029, they will be able to complete tasks that take humans a full working year. That sounds remarkable, but that progress continues to rise. All the underlying trends are moving in the same direction. People often say, “Yeah, but you’re mostly talking about coding, Keith.” But I am not; I am talking about scientific reasoning, maths, robotics, self-driving systems and the use of computers for tasks. In a world where AI is able to complete tasks that would take a human a full working year, we are seeing such a rapid increase in autonomy that any bet that says we need a fully or partially crewed system by—I don’t know—2030 is a fairly large bet against AI progress. So far, everybody who has bet against AI progress has been wrong, whether it was that AI would not beat us at chess in 1997; that AI would not beat us at Go, because of human intuition; that AI would never beat us at poker, because humans are better at bluffing; or that AI would not be able to outreason us—until it did by winning international maths olympiads. Now we are saying that AI will never outperform us in judgment or in controlling vehicles of all types. It seems to me that that is the next thing that will fall, and on those rates—2x every four months—it will fall very quickly. The key question for the Committee is, what time horizon are you considering the future of war against? What does that mean against the rate of progress in the technologies I have described? We are on for something like the Cambrian explosion in scientific advancement and in AI capabilities if the trends continue as they are. You need to come to a view on whether you expect the current trajectories to stop or to continue. What do the UK Government and UK defence think? Any bet in the direction that says, “No, self-driving systems—in this case drones—are going to have only this much impact on the future of war,” is a pretty big bet against the progress we are seeing. My guess, on the balance of probabilities, is that that progress will continue.

Defence reform comes into all that as well: is the MoD ready? Time is limited, so I will move on to my next question. Sir Hew, to what extent might an emphasis on the Ukraine conflict distort our understanding by overshadowing lessons drawn from other theatres of war?

Obviously, there are peculiarities in the Ukraine conflict. Much more important is the point that Ed made, about the tendency, particularly in the United States, to say, “This isn’t how we would fight.” We would of course try to avoid what has become an attritional, protracted conflict, but whether we could avoid it is a rather different question. When Ed talks about lessons learned and the way to address them, there are big handfuls of things to which we failed to pay attention because we concentrate on things like the drone debate. First, Ukraine still does not have an effective system of conscription, and it never thought about it properly before the major invasion. Ukraine thought about lots of things between 2014 and 2022, but it did not address that, and it still does not have a whole-manpower approach to the mobilisation of society. It is suffering dramatically from that, in terms of both the numbers at the frontline and the societal implications of not using manpower—by that, I obviously mean men and women—in ways that actually reflect the skills that individuals have. You do not want an IT expert being a frontline infantryman. Ukraine still very much has a land-dominated headquarters with a land-dominated approach. Ed and Keith are very much involved in the air dimension, but I would point out what is happening in the economic war. If you attend any briefing, people talk about what will happen in 2027—it used to be 2026, and before that it was 2025—in terms of Russia’s economic crisis and how that might affect the war. Equally, of course, there is a point at which Ukraine can no longer sustain the fight either. We sustain by dribbling, but we do not sustain by giving overwhelming superiority. Those are the big-handful questions that we need to address. On the other hand, what Ukraine has got absolutely right—it is now an unfashionable word, and we use other euphemisms—is the propaganda war. It has fought a brilliant propaganda war in getting international support and generating support domestically. Many expected Ukraine’s society to collapse internally because of the effects of corruption and pro-Russian sentiment, which would cause division. So far, it has not. There are tensions there, but Ukraine has dealt with them.

We have only 40-odd minutes and there are huge areas yet to cover. We want to delve deeply into the SDR, the DIS and strengthening the defence industrial base, but before we go into those three areas, we have a couple of quick supplementary questions.

I think Atlantic Bastion is moving in the right direction—large numbers of autonomous, likely attritable systems that can be produced in much shorter timelines, working in an integrated system. That seems a pretty sensible bet. I am not necessarily suggesting you need to go all in on this, but putting 10% of the defence budget into novel technologies would see you transition over time, and position you to transition when you needed to fully. If they do prove obsolete—I am not arguing they will—you would be ready for that, having prepared and hedged against that outcome. Atlantic Bastion is moving in the right direction, but we do not have an equivalent of Atlantic Bastion in either of the other services. Just for wider awareness, Atlantic Bastion is an attempt by, principally, the Royal Navy to have an outcomes-based procurement—“We wish to achieve this; can industry build it?”—and then fund industry to build that solution, and to do so not based on just large capital expenditures on the kinds of thing that Mike is asking about, but with the speed of change that we are currently seeing, so you do not end up with huge cost overruns. I think there are some good signs there, but you need that on a much bigger scale.

Morning, everyone. Dr Keith Dear, just going back to what you said before, is there anyone you can point to in the MoD or Government who understands the pace of the advancements in AI that you have just described?

I think we should be fair and say there are quite a few people. I do not know if I would do them any favours by naming them, but an obvious one would be Jade Leung, the Prime Minister’s AI adviser. Previously, Matt Clifford was very seized of how important all this is, and he was making all those arguments internally. There are other people within DSIT more broadly, and there are a handful of people across the MoD. So there are those people. Are they winning the argument? To some extent, but they are doing it by trying to work through a system that is not really set up to move at anything like the speed or the urgency, and that has a broader dogmatic commitment to perpetual scepticism against AI, which I think is a fundamental problem across Government. Those people do exist. As I say, I am not going to name all of them, because I do not think that would necessarily help, but they do exist—and more power to their elbow. The problem is they are working within a system that is not set up to move at the speed that technology is moving or against the kind of thresholds we are reaching now. As Edward says, it is no longer, “This might come in a few years.” It is here now: fully agentic cyber-attacks run through Anthropic’s cyber-models. Logan Graham, who was my former colleague in No. 10 and the Prime Minister’s AI adviser, who is now at Anthropic, said that this year, Anthropic expects self-improving cyber-physical systems—that is, robots that self-improve as they learn over time. Those things might be building the submarines that Mike refers to. So there are those people. I do not think they have the influence they need to have, and I just do not think the system is set up to cope with the speed of change that we are seeing.

Arguably, the idea of a platform is not compatible with the speed that you are describing. I have a couple of questions related to defence capability and sovereign power. Sir Hew, before we start on that, can I ask you something that picks up on what you said earlier about law? I should declare interests of my own in training and skills in this area. You talked about the rules-based international order. Of course, there are two issues there. One is the process of going to war. In parliamentary terms, that has arguably been very badly compromised by the Iraq war experience, which dipped the hands of parliamentarians in a piece of what had previously been, as it were, prerogative foreign policy, and therefore diminished accountability. I would be grateful if you would comment on that. The second is the question of the use of law and, indeed, the replacement of the law of armed combat by rights-based law, which is asymmetric in its effects and does not necessarily, in this modern world, buy us as much legitimacy. Of course, we think it is the right thing to do, but it has a social impact as well. Sir Hew Strachan: I think part of the problem here is that there is an essential paradox in the whole construction of the international law of war, which is a prohibition on interstate war, except in cases of self-defence, and a desire to enable armed forces to go to war but within a legal framework. That is, at one level, an absurd paradox incapable of solution except in cases of national self-defence. When you go to war, you need to argue, whether you are accounting to Parliament or to a wider international audience through the UN, that you are going for the sake of self-defence. The result is, of course—and Iraq is a case in point—that the argument for self-defence gets stretched beyond the bounds of credibility in order to sustain that argument. The UN Security Council resolution became, at least for the then Government of this country, the be-all and end-all of justifying what we were doing, and it did not come through. Should there be direct parliamentary control of that? My view is no, because a short-term crisis might require the Prime Minister to make an immediate response. Should he be accountable to Parliament for that decision? Of course he should be, but that is the precedent that we have already.

Arguably, that accountability is reduced by having parliamentary votes on these matters.

Yes, I would agree. I think David Cameron had a real problem in—I am losing track of the years—2012 with the whole issue of how to proceed in relation to Syria, which was a problem largely of his own making, in my view. That is one side of the argument. As far as rights-based law is concerned in relation to the law of armed conflict, I think we have two pressures that are, in some ways, disabling the armed forces and the use of armed force. One is a very well-developed system of international law beginning with the 1977 additional protocols covering particularly non-state actors in relation to war, which were largely legacy instruments to deal with post-1945 colonial conflicts that are now applied in other contexts. On top of that, there is the International Criminal Court—of which we are a member and supporter, but the United States is not—which we wish to endorse and support because that absolutely cements our position of being a supporter of international law. But it is jeopardised principally by—much more than by our own use of rights-based law—the United States’ application of rights-based law, which is essentially domestic law being applied in international contexts. There are real challenges here. The much bigger issue is UN reform and how much anybody is up for that in order to make it a more effective instrument. How do you reform the UN General Assembly which, broadly speaking, has voted in directions that you would want to support, when it is not in the interest of great powers or particular members of the Security Council? The other issue is how you align the legitimate use of armed force with the capacity to use that armed force in ways that are effective. The Ottawa convention on landmines is a classic illustration of that dilemma because if you were fighting a defensive war in a land context, you would need to use landmines in the current set-up, as Russia’s immediate neighbours are recognising. You cannot just take a position solely on the basis of what you see as an international agreement to which you have become party if that actually jeopardises your own national sovereignty.

And that might, in fact, increase the likelihood of war.

Absolutely.

Okay, that is incredibly helpful. I am going to reverse the order of questions that the Clerks have written—not that I ever ask those questions anyway—and start with the sovereign defence and sovereign capability question. Dr Dear, you have talked about AGI and intelligence—of course, there is a whole series of equivocations about the word “intelligence”, but let’s leave that; my own view is that “capability” is a more useful word. Can you just set out where you think our sovereign capabilities ought to be and how good we are at thinking about those strategically in the way you would like?

I think there is no upper limit on the amount of intelligence, however you define it, that we might wish to access if we accept that all the things we care about—weapons, tactics, strategies, policies and so forth—are the products of intelligence. If that is true, there is probably no upper limit on the amount of energy that we will need. The more intelligent the models become, the more energy they are going to need. We are seeing 10x improvements in the efficiency of models every 12 months, and we are seeing demand rise as the models become more capable. I think that will continue, particularly if we are in a wartime situation. Energy independence—or at least the ability to scale your energy supply rapidly—will therefore be absolutely crucial. That is probably the most important determinant of where you get to, assuming that allies are willing to share their models with you. Absent that, it would be wonderful to think that the UK could be doing two things at the same time: it could be building models in the way that the big US AI companies are and investing in them at the same scale, and at the same time rushing them into industry in the way that China is. But we do not have the industry, and we do not have the ambition, frankly, for building the models in the UK. Therefore, I think a more sort of spread-betting approach—

It also does not seem as though we have the strategic intelligence to understand that there are different ways of thinking about model construction and that, for example, China is adopting different approaches from America and that there may therefore be a spectrum in which we could play effectively with sovereign ability.

I think that is right. But I think we need to disaggregate the “we”. For example, ARIA’s Nature Computes Better programme might be one of those big bets on the future; it sees biological architectures as the path to much more efficient and effective intelligence, and perhaps as the path to superintelligence—or a path that completely surpasses the in-silicon versions of AI. There certainly are bits of Government—ARIA is a Government-adjacent agency—that think about this in the right way and do see it. But then the question is, do we allocate capital, time, attention and resources proportionate to the opportunity that those kinds of alternative architectures offer? I think the answer to that is no. I think it is the same for compute infrastructure. The arguments that Matt Clifford and others in Government have made for what we need to do are right, but the speed at which we are pursuing them is nowhere near proportionate to the risk, threat and opportunity that we face. The idea that you can completely own the stack is not even true of the US. The chips they rely on are, in turn, reliant on machine tools built by ASML out of the Netherlands. I think everybody accepts that complete sovereignty is not possible. The problem is that that can become too much of a hand-wringing argument for no sovereignty, and I do not think that is good enough. I also think that it dilutes the focus on what we can do, what agency we have and how quickly we need to move.

So there is compute, energy and nuclear. Ed, do you want to add on sovereign capability? I've said nuclear, but maybe there is a question mark on that.

Can I just emphasise something? Spread betting across alternative architectures is something the UK could, and is, doing, but it could do that much more broadly than it currently is if it wished to bet on the future.

Have you written about that?

Yes.

I would be very interested to see that, if you want to send it in.

We need to work out what prototype warfare looks like. We are still oriented around designing things where you have to cut big amounts of steel, design the spec and agree the spec, and then the lawyers will spend 20 years arguing about whether British Aerospace or those around it are building the carrier you agreed and who’s going to change. We need to move to something that the software world has been working on for a long time, which is the minimum viable product: create a minimum viable product and get it out to the frontline. Back to my point, get the services to take minimum viable products, get clever young troops working it out and then send the results back to industry. We need to think through the model of how you do prototype warfare. Then your defence industrial strategy is not picking winners; it is creating the financial conditions and the conditions for industry. Energy supply, energy costs, compute power—those are going to allow these companies to be successful. Government needs to set the conditions and get out of the way. I think what it is trying to do at the moment is shift the model where we define big military plant and the way we build it, and trying reluctantly to move to thinking of information age products in the same way—

It sounds like the defence and security strategies we have are not only failing to acknowledge the point about the substantive requirements, but are just the wrong shape and design to hit the concerns that are being articulated by this panel.

Yes. You are trying to define exactly what you need and then to find a way of building it over the next 10 years, and that is not going to work. In fact, the SDR—I reread it again this morning to make sure that I wasn't misremembering—says that you cannot start defining now, over a 10-year timeframe, exactly what you are going to get in 10 years’ time, but that you need to be building a system that can build it possibly only minutes before you actually need it, given the speed at which design is going.

Sir Hew, do you want to come in on anything?

I would just make one historical point and one alliance point. The historical point is that in wartime—even in the two world wars—the whole procurement process is months. You have this continuous feedback loop from front to rear, and design and adaptation, and we have simply lost any sight of that. The other point would be about spread betting in relation to allies. We are not going to have a fully sovereign capability, so whom do we rely on in the current environment? That is a very difficult question, so you are going to have to spread bets there. My reading of the two US documents is that if we become partners rather than dependants, we are more likely to get what we want than if we do not. But we still will not get it if the US is engaged in a major war with China, because it will not have the capacity to give us what we want because it will want it. That really is the core question. The weakness of the US national security strategy and national defence strategy is that the US does not address its own lack of capabilities for the threats that it thinks it is confronting; it assumes they are there. It stresses economic growth and the defence industrial base, but it does not actually say how it is going to deal with this problem. Without that, how will it be able to supply—as it wants to supply and has supplied—to the United Kingdom if it is also engaged in dealing with its own hemispheric threats?

I have a tiny follow-up question. When a drone company tries to indigenise in this country a drone business that is a sovereign capability of another country, should we regard that as adding to our sovereign capability or as a potential strategic weakness?

It could be either.

Who owns it? Who owns the output? How does it feed in? Part of the problem is that if you generate the IP, it will go overseas, so where is it going to go to?

I just ask the question. Thank you very much.

Keith, how is the absence of the defence investment plan impacting preparations for current and future conflict?

A central focus of the strategic defence review was the digital targeting web. The MoD has a genuinely very capable team put together to lead the work on that. The digital targeting web, as conceived in the SDR, was fundamentally a new operating concept. It goes back to Ed’s point—it is a theory of winning an argument for how we would fight and win future conflicts, with everything flowing back from that. It could not be more central. The SDR set a deadline of a minimum viable product being in existence by this year, I think—certainly by 2027. The absence of that funding and the further delays to the DIP have resulted in the pausing of the forthcoming industry day. It is therefore hard to argue that the progress on that is not stalling. That is just one example, but it is one of the most central, totemic things in the SDR being slowed down by delays. I would suggest there are delays in decision making; I cannot prove the causal link, but it seems pretty obvious to me that that is what the link is. It is having a real and meaningful effect. For all the talk of the need to deploy more capital, we are not seeing any of that reflected in the rapid award of contracts in new areas, which you would expect. The MoD could have a dashboard to show what it has invested in, including the 10% committed to novel technologies, and how much money has been allocated. You could watch it in real time, but none of that exists. That is partly because the MoD just does not think about dashboards in the sense of tracking progress and outcomes publicly, but it is also because the delays mean that progress is not happening.

Ed, do you have anything to add to that? The DIP was supposed to come out in the autumn, then we were told it would be towards the end of the year, and now we are told it will be March. What do you think is holding it up?

I think it is that it is trying to make this 10-year silt chart work. It is trying to predict the future and then allocate moneys over that period. Does everyone understand what I mean by the silt chart the MoD has for all these big programmes? I agree with Keith. I do not know exactly what the causal factor is, but because we do not know where Ajax or some of the other big programmes like the E-7 Wedgetail in the Air Force are going, you cannot agree everything, so you cannot agree anything. To link this to the problems of the industry, it is not just, as Keith said, that things are moving very quickly; these companies have to keep churning or they die. The big cultural void between the tech sector and a lot of the public sector—and I speak for those in uniform and without—is that the latter do not get the need for speed. They do not get that if Keith’s company has a design team, he cannot sit there for three months and then six months waiting for someone to sign it off—he has to move on. One of the key things that the NAD has to do is appreciate the relationship that defence has with the big primes—some of that will not change—and it also has to reconstruct. The problem with the defence investment plan is that it tells me that it will be business as usual as it has been for the last couple of decades. Even though the rhetoric is that we have to work with small and medium enterprises and get into bed with tech, we do not understand what keeps them alive and we might be killing them.

Does future warfare require a different approach to design and production of platforms, given potential attrition rates in a peer conflict?

Yes it does. As I think I said, we are seeing the shift from very complicated manned hunter-killer platforms, of which something like what you saw at Barrow, the SSN, is the ultimate exemplar, to something where you have a much more federated network where every individual platform—however big the warhead—is connected and the thinking glue is somewhere in that cloud making the decisions that Keith referenced. Future warfare does look very different, and your defence industrial base and the supply chains behind it are going to look very different too.

It depends on what you mean by “future warfare”. We have assumed major interstate war. That is the question, but of course, you do not know.

Yes. Where is technology going to take us?

I also think that being able to innovate and develop new capabilities at the speed of technological advances in peacetime and grey zone conflicts is essential to being able to deter effectively. Showing that your industrial base can move that quickly is equally central. At the risk of drawing Sir Hew’s ire by referencing history, which I have tried to avoid since his earlier comment, we can look at the way in which we developed capabilities in the past. Go back far enough and we can see that we invented the weapon then we spent an awful lot of time figuring out how it could be optimally deployed in combat. By the time we got to the first world war, that was beginning to change because scientific advances were such that we could start with the operational problem and then invent the technology to solve it. By the time we got to the 1970s with DARPA’s Assault Breaker, we were thinking concept first. We were looking at the Soviets and saying, “Hang on a minute, there are second echelon forces. How are we going to beat them?” We came up with a plan that was concept first and said, “We’re going to win by having precision-guided missiles, sensors that can see through clouds, aircraft that can fly at a low level, Apaches and drones”—a whole raft of things that now dominate modern warfare. But it was the idea, the thinking or the concept that came first—“How are we going to win?” Then the technology followed. Going back to that US AI strategy—and I cannot say it enough times—they are talking about speed being at the absolute centre of everything and new weapons being created in hours in response to intelligence. In that world, if you try to follow technological trends, you are only going to fall further and further behind the frontier. You have to think concept first—“Where are we going? How do we build for the world or win the fight that is coming?”—rather than just following the technological trends, because that is a certain path to falling further and further behind others who do adopt that approach.

Following on from that, given the length of time it takes to procure this new technology, is there a danger that it is out of date by the time that it gets produced?

It is a danger, and it is a reality in almost every major defence programme that we have.

That is why we mentioned prototype warfare. You do not want to be buying things and putting them on the shelves for them to become obsolete in months.

First, yes—and I am sure that you are preaching to the converted—but secondly, I think there are many in defence who would say, “No, that will never be true.” It would be treated as an absurdity, but I think that it is fine.

I agree, but I think they would say that it might be true of niche capabilities, but not of much wider defence industrial capacity. I think the question at that point is: we want to get beyond this dogmatic commitment to perpetual scepticism, so what would change your mind? When that happened to DeepMind, when people started to say, “All you ever do is win at these games like chess, Go and shogi”—

I think that is true. It is about the feedback loop being there: when you start failing, you start improving, because nobody can deny any more what is happening. But the Committee specifically has a real power here to say what would change its mind. What will convince you that AI will do the things that you described, because then we can set out the challenges to prove otherwise? We can set out the benchmarks that say, “If a model could do this, we would believe that this is possible.” I think you have a real power to change that.

Air Marshal, do you want to come back on that briefly?

How would you test that? Many years ago, both Keith and I were involved in a technology demonstrator programme to build the virtual world, and we used exactly your arguments. We said, “If you can build the virtual twin and test it”—

That was one, but other companies are available. In your virtual world, the real-world intelligence gets put in new models with lots of contingencies. The science-fiction element was that eventually we would just hit print, and the 3D printer would print the weapons off when we need them. I do not think we are quite there yet, but we are getting closer and closer to that.

Last year’s SDR talked about a whole-of-society approach to bolster the UK’s resilience. How will moving in that way have an impact on the challenges we face for future warfare?

I have spent quite a lot of time thinking about this. We need to move beyond talking to doing. There has been a lot of talk about how we get a whole-of-society response and think about national resilience. We failed to build on the experience of covid-19, where there was a whole-of-society response to the threat.

Do you think there was a whole-of-society response?

There was a local community effort, which, in my experience, was generated from the bottom up, not from the top down. We have simply failed to bottle that, in terms of where we could put that elsewhere in comparable contexts. Much of the issue about cyber-security is about people obeying basic disciplines themselves in the workplace, which they do not do. It is far too complicated to change your passwords regularly and engage with that sort of approach. There is an element of the bottom-up approach that is about having a dialogue—going back to my answer on deterrence—with people in order to help them understand how they can contribute, instead of mocking what is going on. Specifically in relation to the SDR, we are nowhere near having any concept of how we would generate mass and how we would replace the existing armed forces. We have talked a lot about what technology will do, but if we look at casualty rates in the war in Ukraine, we have no idea how we would confront that. We do not want to have a debate on conscription, because it is too difficult, so we go round the houses looking at other alternatives, which imply some element of voluntary engagement.

When we visited Finland, we looked at their whole-of-society approach, which is utterly different.

Indeed, Finland has a totally different response, but it has a cultural awareness, a historical precedent and a discussion going on the whole time. We are self-evidently not in that position. Let’s take the Scandinavian countries—Finland, Sweden, Denmark and Norway are now thinking collaboratively about how they do this—and how different their experiences are. For example, Norway sees its threat as being to the north, but its population base is in the south. How does it actually get a conscripted army to be in the right place at the right time? How do you think about where the contribution is of the IT specialist who also happens to be a service reservist; when mobilised, is the priority to remain in the IT job or to go and fulfil the service reservist job? You need to think that one through in advance. What allowances are you making financially, either to the employer or to the family of the person concerned, particularly if there is a drop in salary as a result of fulfilling the reservist obligation? To look at the future, one of the commitments in the SDR is to think about increasing the cadet force. If you do not have any adult volunteers—and at the moment there is a massive shortage of adult volunteers in the cadet force—you will not have an increase in the cadet force. Are you going to pay your adults to be volunteers, or are you going to think of another solution to the problem?

If we are going to go down that line, I should put on record that my son is a volunteer in the cadet force. We are not advocating for them to be paid.

There are some wonderful adult volunteer cadets. I was Lord Lieutenant of my county for 10 years, and I got to see a lot of them in action. Yes, of course, but the point is there are not enough. My issue is not that we do not recognise the need for national resilience or do not have the discussion—we do. What we have not done is move to the point, specifically within the MoD, about how we think about that and how we replace the armed forces that are currently in existence, on the assumption that they will not survive for protracted conflict—not, I am afraid, in their entirety. We are talking about lethality and attrition in current warfare—that carries implications. We need to think about how we replace those people. In wider societal terms, we are not talking about the practical steps that we would need to take to enable that to happen. We tell people they need to have bottled water and generators at home, and all those things are good advice, but it goes beyond that.

What should be in the defence readiness Bill?

What are the lessons from Ukraine and the applied element of that? Do as you have done, go to Finland and other places and demonstrate how a resilient society converts into a confident society and a vibrant economy. Do what is in the Government’s defence industrial strategy and say, “Let’s talk about making this an element of growth.” Some are risks that only Government can take. If it is applied to companies like Keith’s, it would be dual use. Sweden and Finland have very good fifth-generation IT companies. After that, things such as thinking about the mobilisation of the reserve and how to answer the questions that Hew put forward. You need to consider the fact that a lot of our critical national infrastructure is now in private hands, especially in the digital sphere—data centres, etc. What are the critical industries and how would you take them over? As Keith has pointed out, the American Administration has said, “If your AI becomes so critical, we can nationalise it.” We have got out of the business of civil defence. We need to think through the denial part of deterrence. I would just throw those in as openers.

I would offer a couple of things. First and foremost, we need a war book on the scale we had before. It is what we do, not what we say that matters most around that, to enable that national conversation. Seeing that the Government was making every Department, the BBC and others have a plan for how we would respond in the event of war would show that we were serious and enable a much more grown-up conversation around what that means. The defence readiness Bill would be really clear in demanding a war book and the level of detail required in it, to be something approaching what we have had previously. It would commit to two things that Senator Warner did in about 2010 in the US. Back then, I think it was the House Armed Services Committee that managed to push into legislation that the US had to have 50% of all frontline capabilities fully autonomous by 2015, if I recall correctly. Most of the capabilities we have today were driven by that mandate. They had no choice; they were forced to do it. I was on exchange with the USAF at the time. They hated it; they had to do it. Eventually, they got it unwound, which is why we have not gone as far as we might. That defence readiness Bill for me would define what AI first means. It would define what novel technologies are and it would make a demand for much greater autonomy to a level of percentage of frontline capabilities. All those things would be essential if we are going to have any chance of moving at the speed we need to.

I would say that just-in-time logistics have to be addressed. I was staggered to read that we still produced 60% of our own food the other day, which is the same as it was in 1914 and we thought we were not ready, with six weeks’ supply in the country. We need to think about that. We need in the same vein to think about the relationship between this and issues such as war insurance. Do we have a war insurance scheme in place? We need to have discussions, which stopped in about the 1960s, between the Navy and the merchant tonnage on which we rely. It is extraordinarily complicated now with flags of convenience, shipping regulations and all the rest of it. We focus again and again on fighting power and the land dimension of that, which I say in the presence of two airmen. There is a great deal of attention on that because that is our image from the past. The image from the past that we actually need to think about is how we do this economically over the long term. It seems to me that we have not begun to engage with that discussion.

Judging by that, it will be quite a large Bill with quite a price ticket attached. How can you build into legislation provisions that keep pace with rapid changes in the ways of war, with artificial intelligence? You referred to the war book, but we are in a different time now. How would you legislate for events that are going to take over that legislation? What do we do there?

I will just quickly say that a plan is better than no plan. It may not be the right one at the time, and you will almost certainly have to adapt it because the circumstances will not be fully as you have anticipated—there will not be perfection on this. It is a bit like the procurement argument. If you do not think about this at pace, and do not have a risk appetite, you will not get there. It will not be perfect, but if you do not even think about it, you have got nothing.

There are two other things. One is organisational design: we do not have a system that is designed to sense and respond rapidly to threats. Everything is designed to slow things down in Whitehall, rather than to get things done. It would start with a machinery of government redesign, specifically on our national security, as opposed to primarily our defence architectures—defence would come second. That would be essential. Secondly—I would say this, wouldn’t I, given what my business does—we have a science of forecasting and prediction now. You would apply that at scale everywhere, because everything is a prediction. A decision is a prediction: “If I do this, I expect this to be the outcome.” Tactics and strategies are a form of prediction: “If we do this, we expect this to happen.” Resilience is a prediction: “We need to invest here in order to avoid this outcome.” Today, we have models—we deploy what is genuinely the world’s leading model in this domain, ahead of Grok, OpenAI and DeepMind—that perform at superforecaster levels in predicting the future. You would apply AI and collective intelligence tools to leverage the best of humans, and do so continuously, in real time, to anticipate what you need to respond to, and in turn to predict what you should do to respond to maximally mitigate any threat. I do not think that thinking exists in Government at the moment; we still think about planning as something that is done by a group of humans who are sat around a table having an argument about what needs to be done, and not about the application of the science of forecasting under uncertainty and the increased use of AI to do that much more efficiently and effectively.

Do you think such a Bill is needed, or should Government just get on with doing what they need to?

I think without a Bill, much of this will not happen.

I agree with that completely.

One thing I would say positively—and I do not know how far it has even been responded to—is that people have talked from time to time about the need for an incoming Government to go through a wargame exercise to actually experience what it would be like and to come up to speed with it. Memorably, New Zealand tried it once. It happened on a Saturday afternoon—perhaps the All Blacks were playing a match—and the pilots who were designated to get people had drunk rather a lot and could not respond, because the day that they chose to do it was of course a day when people were not expecting it. Just getting people sensitised to the issues—that is all we are talking about—would be a starting point.

We are coming to the end of our session. I will bring in Mike Martin, and then Derek Twigg.

No.

Everywhere I go at the moment—not just in defence, but in the think-tank world that was referred to at the start—within five minutes, the discussion comes back to the thicket of process. Not just that of Whitehall, but of all the regulators around it and the judicial reviews. A fundamental problem at the moment is with the ability to get anything done. Unless that is addressed, short of a Cuban missile crisis-level shock, we are still at the moment churning the old processes and hoping for different outcomes.

I think proportionally it is still much less likely that we get into a major conventional war, probabilistically, than that we do not. It is important that we do not over-hype the threat and that we look at how likely it is. The likelihood of it happening is increasing, without becoming greater than 50%. What we are not seeing—

You are very welcome to commission Cassi to do that, but I offer you my personal view: we are probably at around 20% probability now, where a few years ago we would have been at around 10%, depending on how far back you want to go. There is where we were in 2014; now we are at around 20%. That is a much bigger risk, we ought to be investing proportionately to it, and I do not think that we are. That is my concern. That is not to say that we are more likely than not to get into a major conventional war; I do not think that that is the case, but unless you prepare you will see that probability increase over time—I think that is the point that we are all making.

I think that you have covered them. It is worth highlighting intelligence. That is where Europe really has to get its act together. The problem is, to go right back to the first answer, if you have removed America from that western bloc, you have removed the team coach, the team captain and a lot of the backroom staff. It is very difficult to see who in Europe would step up. If the Brits step up, you are not in the EU any more because of Brexit. If the French step up, the Germans will be concerned. The French will not want the Germans to run it. Everyone was happy to play second fiddle and let America dictate, so how do we organise 600 million west Europeans to deliver the clout that their economic latency suggests they should be able to deliver? I would start with intelligence: how do you replace the Five Eyes intelligence that is driven by American collect? You then might start to flesh out some of the other elements. Within the national sovereign base, I think that you have covered them all and there is nothing that immediately leaps out.

I would like to see us invest a bit more in the space environment. We have talked quite a good game there and we are good at the apps around it, but core space capacity is another area where we could invest more.

I would start by echoing space. Space will be the most lucrative market in human history. It is probably going to be the primary determinant of your relative power in the international system. I think that we have no conception of its central import to our relative power in the future. I would like to see much more investment in a really serious space programme in the UK. I know how that makes me sound and I still think that we should do it; I am happy to put that on record. We have real leadership in quantum computing because successive Governments were investing quite early on. We had a quantum computing strategy long before other nations. As ever, that leadership is in danger of being squandered when it comes to how we start to commercialise it and scale it. We need a focus on how we pull through and accelerate an advantage that we have had, and I think still retain. I also think that biology and specifically synthetic biology are going to have profound effects that are hard to predict. The idea that you are going to be increasingly able to design biological systems, adapt biological systems and indeed combine those in cyber-physical systems with the wider technological advances that we have talked about is going to have profound impacts on our prosperity, our security, and how we will live in the future. I was told consistently in Government that we have real leadership in that area, and I am now told—such as when I talked to people outside this session—that we are again not pulling through and in danger of losing, as our pharmaceutical companies and others perhaps look elsewhere. Those are the areas that I would point to. Biology is the one that I know least about and therefore worry about the most. In the other two areas we could do an awful lot; we just need to understand the profundity of what they offer.

I have two quick points. Intellectual rearmament—in Ed’s phrase—is crucially important. That is what we have come back to again and again. I think it is a sovereign capability and we lost sight of it. We could do that and it does not cost much. The other thing I would say is: do not make sovereignty such an obsession that it gets in the way. The Scandinavian countries are now thinking about Scandinavia as a joint airspace for interoperability. They are maximising effects rather than separating effects. That is a capability more than, if you like, a deliverable in terms of the raw technologies, but the point is that we are not going to be able to do it on our own, if we are talking about a major war. There may be a low probability of that, and a counter-terrorism thing may actually be the immediate issue that we have to confront—the sort of thing that we have not talked about but which we would have talked about 10 or 20 years ago. One question that we might have, and that we have not debated, is about how far you are disabling yourself from dealing with some other issues that fall within the defence space because you are looking at the biggest issue. That was always the cold war problem.

This has been a very thought-provoking, fascinating and sobering session on the nature of future warfare. A lot of ground has been covered in the last couple of hours. However, there will no doubt have been other areas which have not been covered. As I said to the panel outside the room before we began this session, I welcome either collective or individual submissions from you that you think could inform the work of the Committee. I thank all of you so much for lending your expertise to this session. I am sure that the wider British public will have gained a great deal from it, especially with regards to enhancing the national conversation. With that, I bring this evidence session to an end.