Welcome to this final panel in today’s hearings on the US-UK investment and trade relationship. Thank you very much indeed to our witnesses for joining us. Could you introduce yourselves and your companies? Then we will get into the questions.

I am executive director for UKQuantum, the association for the UK’s quantum industry.

I am the vice-president of strategy for CoreWeave and managing director of CoreWeave International. We are an AI-based cloud business that has a software stack optimised for running AI workloads. We have a number of data centres in operation across northern Europe and the US. We currently have two in the UK and have just committed to invest and build a third one in Scotland in the AI growth zone.

I am vice-president of corporate affairs for Microsoft, a tech company, here in the UK.

Hugh, let us kick off with you. How significant was the technology prosperity deal in unlocking the Microsoft investment announced during and around the state visit?

It was helpful in focusing our attention. We have been in the UK for nearly 45 years and over that period have made a range of different investments to establish bigger offices, have more capacity, hire more people and build Microsoft Research in Cambridge. More recently, we invested in Microsoft AI with Mustafa Suleyman based here in London. As we build out data centre capacity, the investments continue. Around four years ago, we invested about £2.5 billion. Last September or October, we made an additional investment of about £23 billion. The deal was helpful in focusing attention on the question: “What does this look like in a more organised fashion?” We look at our investment path in terms of both opex and capex. How do we organise this in order to drive efficiencies? How do we deliver for the Government in terms of what they are looking for? Essentially, we build Lego blocks for others to turn into AI services. With the technical prosperity deal, the focus was on prosperity. We were very pleased to get behind it, as it unlocks a number of things that are quite exciting for us. But there are other aspects of Britain’s investment climate that are perhaps more effective in unlocking more investment.

Let me come back to that last point. What was the bait within the technology prosperity deal that made Microsoft think, “Ah, right, this makes a yes decision much easier.”?

There are a number of things. One was the focus on quantum, which I imagine you are going to get on to. There is a tremendous opportunity for the UK, which has some of the world’s greatest quantum capabilities. If a deal can help orchestrate and smooth the lines of communication between the US and the UK, and build out that opportunity, there are good, positive prospects for the UK. Another was the alignment around security. The UK came out of the blocks very fast on AISI, which was originally called the AI Safety Institute and was then rebadged as security. If a prosperity deal can help create global alignment around those sorts of principles, particularly between the UK and the US, that is all to the benefit of AI and the security and safety of AI that operates here in the UK and elsewhere.

That is very helpful. We will come back to the difference between what was in the deal and the fundamentals, because it is helpful for us to understand this. Ben, what seduced you in the deal?

Hugh used the word “helpful.” It was a helpful backdrop for an investment decision. When we are making investment decisions on the international business, we are looking across the whole of Europe and asking, “Where do I want to allocate my capital?” A great advantage of the announcement was it gave a degree of certainty and comfort around policy and the environment for investment within the UK. In the earlier session, there was discussion about other requirements, such as planning and the availability of power. These aspects all feed into the investment decision. But it is very helpful to have that in the backdrop, to know and see that there is Government support, and importantly, to see that follow through over the following six to 12 months.

I do not know how you make your investment decisions, whether through a board or investment committee, but give us a flavour of the conversation you had with your colleagues when this deal was looming on the horizon and you had to come to a decision point.

We had made lots of decisions prior to that. We had already invested to build two data centres, one in Crawley and one in London Docklands. The announcement then gave reassurance and comfort that, in terms of the third investment decision, we will continue to have the support of the Government. They have set the environment and the policy framework and, as was said earlier, that knowledge that there is the rule of law and things will operate and continue to do so in a stable manner.

But when you look at, say, a spreadsheet to see how you weight risk with project decisions, what is the risk assessment that moves to get this project over whatever IRR or hurdle rates you use?

In relation to the Government?

Yes.

I can go through how we make the overall decision, but essentially, as I said, it is knowing that we have the environment that tells us that our 15-year investment in the UK is and will continue to be supported by Government.

What impact did the deal have on the investment environment in technology?

It was very well received within the UK and US quantum industries. There was no deal as such. It was, in fact, a list of actions and things to engage, which is what we hoped for. For a start, it recognised the importance of quantum on the global stage, and secondly it established the importance of a strategic partnership between the US and the UK. There were then a number of critical things to the industry that, in effect, gave a mandate to go and sort out. There was a long list of specifics, but the key ones were around benchmarking and standards, which are both critical to the industry and a real opportunity for the UK and the US to lead what has to be a global initiative, but which is one that requires leadership. Working with my counterparts in the Quantum Economic Development Consortium was an opportunity to pool our efforts and accelerate the adoption of quantum technologies and start to look at trade policies so that we have a unified voice across the two countries, and to do other related activities. There is a lot of positive support for the UK to get involved with the US in a number of important areas.

The infrastructure we build in the UK is not on wheels. We invest here, and the data centres will stay and bring capacity to the UK. You are right: the geopolitical situation at the moment is more volatile than it has been. We saw that in Davos, and Mark Carney’s speech was illustrative of that. At the same time, Microsoft has invested in the UK for nearly 45 years. The UK is probably our second most important market. The US Government have a self-interest to continue to export their services, and US technology is one of their greatest exports. I do not share your concern about what I assume you are alluding to, which is the continuation of investment and digital services and our ability to continue to make these investments in the UK without fear that they will be disrupted.

The sector is emerging, and is undoubtedly still a global endeavour. We still require partnerships and engagements with other countries. We have to be cognisant of what is going on, but also believe these are short-term things. Industry itself keeps an eye on the situation, but we understand that we have to work together to realise the benefits and bring these technologies to market. At the moment, we watch over them, but we continue to work together at the industrial level.

When it comes to the US, there is an important distinction between corporate investment and national strategy, but the Council of Economic Advisers’ report on “Artificial Intelligence and the Great Divergence” has bugged me since I saw it. It outlines a far more competitive than collaborative future for American artificial intelligence, and this feels like a golden thread that will run through US technology policy, which the UK needs to take quite seriously. Is that on your radars? It might be something that Microsoft has to think about. Am I over-worrying about that sort of thing? What could we do to make sure that the UK does not fall foul of that?

It depends on what we want and what we go for as an economy and as a country. If you look at the lessons from previous industrial revolutions or transformative technologies, quite often the benefits flow to economies that put technology to work on behalf of the economy fastest rather than the inventors of the technologies. The UK economy is well positioned to adopt AI very rapidly, put it in service of the British economy and focus on that as the goal, as opposed to trying to replicate infrastructure, which is very expensive, and models and everything that goes with them. Maybe that is where the opportunity lies. There is a great opportunity for the UK to be a leader in terms of AI and an exporter of AI services. Perhaps a vision for the UK is to be a place where it is easier to set up, build, grow and then export an AI company and services than anywhere else. We produced a diffusion index, which shows the extent to which AI is diffused across different economies. We looked globally at 100 or so countries. In the first half of 2025, the UK was at number eight; it slipped to number nine, but only fractionally and is vying with the Netherlands. There will be a series of policy levers that the UK Government can pull to improve even further the extent to which AI is diffused across the British economy. That will drive economic value that puts the UK in a good position over the next 10 to 20 years.

I am delighted that we will run an entire inquiry on this later on.

We are bookended, in that we missed the boat on proprietary sovereign AI. That is where politicians are mistaken with the idea that we can retake it because we have quantum coming up.

We will leap back to sovereign capability and choke points in a second.

BritishAmerican Business, which we heard from earlier, criticised the technology prosperity deal for lacking core digital trade provisions, for example on data governance or cross-border data flows. How important are digital trade rules to your investment in the UK?

It is possibly not directly relevant, but at the moment we think about the supply chain as components into systems rather than the flow of data. When we get into the provision of network-based quantum computing services, we may well start to think about the flow of data as it is accessed. But that is not dissimilar to the current AI and regular data centre services, so it is not directly relevant to quantum at this point.

We have a series of existing regulations in place around data. As a UK subsidiary of a US business, we comply with those requirements because we need to. We handle less personal data and less data. We are more of a platform people can build upon. To use the example of Lego blocks, we enable people to build the applications of the future. That is a key part of what we are trying to achieve. Regulations around data are managed separately.

I would say they are crucial. The UK-EU adequacy and sharing agreements are crucial to our ability to operate effectively. Our customers rely on them for processing and for a whole range of technology services. If those agreements fall—we have a good length of time to go on the EU-UK agreement, as it has just been renewed—we fall back on standard contractual clauses that allow us to process and move data on behalf of our customers. These agreements make it much easier, cheaper and much more straightforward. This will be even more important as we move ahead. More AI services are coming online. Microsoft Azure gives access to around 10,000, maybe more, AI services to customers, of which 1,800 are currently available from a model that is based in the UK, and so our customers can use all those other services from data centres around the world. As the number and the ability to access a range of very niche and bespoke services increase, those will be geographically placed and customers in the UK will want to access and use them. Users will want to use apps and services in different parts of the world, not just when they travel but for a range of different functions. Those data adequacy agreements are essential to enable that to be done in a seamless way.

Building on that, a feature of recent negotiations around technology with the US has been the concern, if not always the actuality, that other US digital priorities such as seeking changes to the digital services tax or the Online Safety Act, may come into the conversation with regard to the United Kingdom. Would you support such changes, Mr Milward?

I am not aware of those conversations when it comes to Microsoft. For a long time, we have advocated for stronger regulation, including in digital safety. We are very supportive of the work of the AISI. In fact, we have gone as far as to write a draft AI Bill with a series of sections that we think are sensible approaches to AI regulation. We do not shy away from those sorts of conversations. To be successful, we want our customers to trust the technology services they use. If that requires regulation for them to have a higher level of trust, then we are very supportive of that.

Yet the US Administration often seem to consider sovereign regulation by countries such as the UK, which has a different conception of how online services should be regulated and the priorities around safety and freedom of speech, to be anti-competitive or in some ways an attack on US digital exporters. Is that the position of Microsoft?

No, we are very keen to make sure we act in the best interests of our customers. That means we pay our way but also encourage sensible and appropriate regulation that gives confidence to our customers to use the services we provide.

I am a guest at this Committee today from the Science, Innovation and Technology Committee. I would like to dig a little into what you have invested in here, Mr Milward. You mentioned a couple of data centres. I presume these all run Microsoft AI models and services. How much space is there for UK firms to develop new AI solutions? Or are they simply putting apps on top of Microsoft solutions?

Most of the operational layer of our data centres is open source. You could think about it in terms of having silicon and then an operating system for a data centre. That will be our proprietary software, as it were. But sitting on top of that are workloads from customers all around the world—around the UK for sure—running what they want to run on it. If a customer decides they need to spin up a virtual machine in order to have additional capacity to do something, there is a range of options as to the kind of virtual machine they spin up, whether that is open source, which is frequently the model, or Microsoft proprietary or any other. We run a range of different models. I mentioned the 10,000-plus AI services in our data centre that are available to customers. These come essentially through a marketplace of hundreds of different providers. It is a long way from being just Microsoft. The idea is to have partnerships with a range of different suppliers of technology services to supply what our customers are looking for.

So it would be more akin to an Amazon Web Services than—I am trying to think—perhaps Windows 95, but that is just showing my age.

Maybe we should make a distinction. Azure, Microsoft’s cloud, is broadly akin to AWS’s cloud and provides a whole range of services that are not just proprietarily ours. Microsoft 365, our productivity suite, is our own. Dynamics 365 is our CRM cloud, and that is our own. But Azure is where all these AI services sit and is broadly capable of supporting thousands of different types of models and services.

Mr Richardson, you talked about having Lego blocks for AI servers. What is it that you have produced in terms of these facilities? What are these data centres?

We build GPU-based data centres that run on the latest technology. Unlike my fellow panellist, we have a CoreWeave orchestration layer that sits on top and enables people to come in and start building models. We are entirely optimised for building and running large language models. That is where the technology lies. It is not that we do not have a productivity suite. It is entirely there for either enterprises or what we call digital natives, so AI native companies, to come on and start to build and run those models on our infrastructure.

Are these facilities for them to build their own AI models adapted to their own things and train them up appropriately?

Correct.

That is interesting. Thank you. Mr Legh-Smith, in the world of quantum we are in today, how can we prevent ourselves from going down the usual route and ending up with the doom valley where we have start-ups aplenty, but as my colleagues suggested earlier, as soon as they become big enough to scale up, we lose them to American owners?

The first step is to ensure they are successful in getting their next rounds of funding. A lot of our leading companies are now out there looking for growth-stage funding. The previous panel noted it can be difficult in the UK, not necessarily in terms of the availability of money, but in the case of quantum, it is the availability of the expertise to lead in those sorts of investments. That is why I usually say that the first step is to invest in specialist quantum funds because these are capable of investing in our start-ups. As companies are successful in getting together growth-stage funding, that will keep them in the UK. All the CEOs I know are keen to stay here. If we can see them through that phase, and back it up with a strong national programme—I have to say, we still wait to hear that announcement, but I am assured it will be soon—it will stand us in good stead for a few years.

We will come back to Mr Aldridge and sovereign capability and the risk of American choke points in a moment, but you all alluded to the importance of the technology prosperity deal and the relationship that it symbolises between the UK and the US. It is suspended at the moment and not moving forward. Are there elements of the implementation plan that went with it that you need His Majesty’s Government to move forward with? Or are you supremely relaxed about this state of affairs at the moment?

Our investment continues irrespective of the technology prosperity deal. There are aspects of the discussions that go from Administration to Administration that are quite useful to continue. The AISI work is an example of that.

Likewise, our investment decision is made. We will continue it and we are committed to it.

I talked about two things: benchmarking and standards. These happen between two government-oriented organisations: NPL on our side and NIST on the US side. Because it is a formal policy, that work is difficult to move forward because people are instructed not to engage. On the industry engagement side though, I have a very good relationship with my counterpart. We have developed proposals and plans that we will put into force as soon as we have the green light. There is an opportunity for the Government to get behind those unilaterally, and we can start those collaborative activities now.

Is the implication, then, that the Government need to recognise that the TPD will not be reactivated any time soon, for the sake of argument, and they just need to get clear about what they will do anyway? Would that help you?

If there is a feeling it will not be any time soon, we need to get on to the next plan because there are good elements there. Certainly from an industry point of view, we can work industry to industry. Perhaps the standards would require a little sidestepping and manoeuvring, but there is still work that can be done.

This Committee’s report on economic security pointed to a rapidly increasing level of private sector ownership of public risk. That was a powerful concept that I probably spend too much time thinking about, but it reminds me of how many of my constituents, if I talk to them about anything to do with artificial intelligence and American companies, are not very positive, shall we say. A number of my colleagues in the Commons in particular are also concerned. Hugh and Ben, you represent two companies, but they are probably not ones that lots of my colleagues are worried about. I would like to have seen others that are not here. We need reassurances and safeguards, and as elected officials the ability to go out with confidence and say, “This is okay. We trust this.” What safeguards are in place to ensure security of public sector digital infrastructure? Because that is the nub of it; people do not trust it at the moment.

As a company, I would like us to stay curious and try to understand what people are concerned about.

Let me put it like this then. For anyone who has read “Chokepoints”—or indeed any book about economic security published in the last year and a half—it is very good at lighting up the choke points that the United States Government now control. There are an awful lot of data centres in Virginia running cloud services that are now mission-critical to bits of the public realm or indeed the private sector here in the UK. The United States therefore has considerable power to control those choke points to our detriment, if it so chooses. How do we de-risk that for the UK?

Part of the work is around making sure we have the infrastructure here in the UK. First, that does not need to be built by the Government or private companies, but Microsoft, CoreWeave and others need to invest in UK infrastructure to make sure those workloads can be provisioned more locally. Secondly, if there is worry about the integrity or security of the data that customers such as the UK Government put on to the cloud, that security is the same wherever it is. There is no requirement for local provision, as if somehow being on British shores is safer or more protected. There is security and more can be put on top of it, including customers’ own encryption that could be added to ensure that integrity is maintained. In many respects, it comes back to where the economic interest lies. At the moment, technology is still one of the US’s greatest exports and the UK is a long-term and long-standing ally to the US. There is very little argument to say that it is a good idea to cut off an ally and a source of economic growth and productivity back to the US.

They were threatening to invade Greenland.

The difference we have now is between the political, the practical and the policy. This has opened up fissures that we did not contemplate before. It is about how we, as elected officials, have the confidence to go back to our colleagues and constituents. It is a work in progress and we all have work to do, but maybe it is something we can explore in the AI inquiry a bit more.

The point is that today’s weapon of choice in the White House is tariffs, but who is to say that it will not be suspending cloud services in the future? Part of the answer is ensuring we have a degree of sovereign capability, as you have alluded to, but there are other perils that we worry about de-risking.

Companies such as Microsoft are busy investing to ensure there is as much resilience as is conceivably possible for customers here in the UK. That includes the physical side. For different types of workload with different levels of dependency there are different types of provision, including Azure Local, a service that can run disconnectedly over a period of time for certain types of workload. Equally, we are prepared to put legal protections in place, which include a willingness to defend our Government customers in US courts all the way to the extreme end. We have made commitments in our contracts to support our customers wherever necessary. We are building a combination of legal and governance protections. Our EU customers are particularly concerned about this. Our response has been to build out a new governance structure to ensure a level of scrutiny in the EU. There is an option to bring the UK into that as well, if that is desirable. We can do that. We are already building out those services, as are other companies.

If we started willingly cutting off our customers, we would not have a business for very long. In this instance, there was a US Government sanction against a single individual. We worked with the ICC throughout the whole period, and it was fully aware of all the discussions taking place between us, the US Government and itself in this matter. It was the ICC’s decision to terminate and respond to that sanction, not Microsoft’s. So we were not put in a position where we terminated the agreement; we worked with the ICC throughout.

It is exactly the same point, because when he was sanctioned, he was cut off, and that is because you did that. It was not because he suddenly said, “I’ll stop using my Microsoft account.” There is no reason why he would have done that. So having been sanctioned, you had to follow and he was cut off.

We did, and the ICC has the ability to switch on and off their own users. It is the same in Parliament. The Parliamentary Digital Service decides whether you can operate your email. In the same way, the ICC makes the decision whether it continues to provide service to one of its employees in response to the US sanction.

Let us draw out the implications with Sarah Edwards.

We heard in the last panel that a reason to invest in the UK is that we have a strong rule of law, and that there is quite a lot of investment going on. Blackstone said it was investing $10 billion in data centres, which is great and obviously supports the notion that we will go forwards with AI. The AI action plan talks about the fact that we have sectoral regulators at the moment, and that if that does not work, they will look at setting up a singular body to look at regulation of AI and data. Are there any changes that would make the landscape more favourable for you, given that is the current one we have?

Our view about the way in which AI is regulated in the UK is pretty positive. Pausing a little rather than running headlong into AI regulation has served the UK extremely well. Other jurisdictions not far away from here regret the fact that they went so fast into regulating, which has created real problems for international and domestic AI companies.

When you say near to here, are you talking about the European approach that some say is stifling innovation?

Yes, I am. There is a risk there of a bureaucratic approach that ties companies up and slows them down rather than speeding things up. The UK’s approach has been very sensible in this regard, particularly with its focus on a balance between model and use case regulation. That distinction is important. Model regulation is around how to ensure that customers using an AI model understand its capabilities and dependencies, what it is optimised for, and various other aspects, and that is produced in a way that is consistent and usable by customers. Those kinds of measures are very sensible. On the use case side, it is about regulating the way in which an AI service is deployed as opposed to regulating the AI itself, if that makes sense. You can use facial recognition technologies for a whole variety of different purposes. Some just switch on your computer; others may be used in defence for target acquisition. You want to regulate the use of one technology quite heavily, but the other you should allow for a whole variety of different uses. The sensible approach is to get away from regulating the service itself and the endless cycle of how to define and name it and instead focus on the use case. We will see what happens when we get to what the practicalities of AI regulation look like. I understand there is a desire not to focus on a single Bill but to look at regulation across multiple sectors. If the UK continues with this approach, it is going in the right direction.

Thank you. Ben, did you want to add anything to that?

I would agree with that analysis. When you deal with an industry that moves as quickly as ours, that is not a reason not to regulate but rather to ensure you regulate in the correct way. The pace that we see every day, and the amount of demand and drive we get from our customers, which is really important, means you need to look at it on a use case as opposed to a generic basis because things will slow down. That is an important point.

Do you have anything to add on that point, Jonathan?

We are at a slightly earlier stage than AI, but our industry is very encouraged by the Government’s approach. They have started to explore the matter and have taken a very pro-innovation view on this. There is a forum for established regulators to look at the impact of quantum, but it is on a use-case basis as Mr Milward said, which is the way we would like it, to look at its use and see if there is anything different to the existing regimes. That work is under way and we are engaged in it. The FCA is probably leading, but I was listening to the Civil Aviation Authority’s views as well. As well as supporting the regulators, we have kicked off a Responsible Quantum Industry Forum. AI and ethics are an issue, and we are trying to get ahead of that to understand what it might mean in terms of the use of quantum. There is a collaboration between UKQuantum, the National Quantum Computing Centre and techUK, so very much an industry-wide look at how we lean into those matters. We are trying to get in early to find the best approach.

Have you asked the Government to change any elements of their approach? You alluded to looking at and encouraging the deployment. Are there other aspects to those conversations or things that you have said to Government in terms of, “This is the kind of movement that we’d need, or this is what we’d like to see.”?

On quantum?

No, on AI and data regulation, but obviously it does include the future.

We have encouraged Government to stick with their principles in how they think about the matter and to regulate rather than not to, but to do so along the same lines. We have also had conversations with them on quantum where we encouraged them to invest in a more focused way rather than letting 1,000 flowers bloom with a series of more academic analyses. Our view is that quantum is a great opportunity that we do not want to miss. You could argue there are parts of the AI stack that the UK has missed. The country is extremely well placed to benefit from quantum, and now is the point at which to make the right decisions.

Can I just check? It sounds like the regulatory approach that the UK is pursuing maximises compatibility with that of the United States. Is that a fair summary?

Actually, it sits between the US and the EU approach.

Does it jeopardise US-UK trade and investment in this space?

It has not so far. We look for a balance that does not go too far down an EU-style approach that is much tougher, much more bureaucratic and much more focused on the wrong kinds of transparency, for example. At the same time, the US approach is perhaps a little lighter touch. For the reasons that Mr Aldridge mentioned earlier about driving trust in users, a level of regulation is required, and we support this.

You heard our anxieties about the American potential to misuse the power it has at the moment, but equally we want to define this narrow corridor where trade and investment between the US and the UK is maximised. Do you all feel that we are in that zone at the moment?

Not entirely. To be honest we are not clear what the zone is in quantum. Just before December, I gave evidence to the Science, Innovation and Technology Committee and after a number of questions I had to admit I was not entirely sure what we wanted sovereignty to mean in terms of quantum. It was at that point the Chair admitted that they too had a number of different interpretations of what it might mean. My recommendation then, as it is now, is that industry and Government need to sit down and work through the situation to see the opportunities and what boundaries we need to put in place, because I do not think industry can work this out naturally by itself.

Do you have anything to add to that?

In relation to our investment decisions and the fact that we made them prior to the arrangement being in place, it is not a sovereignty point but goes to how you make decisions and progress between the two jurisdictions.

Last November, I read the Tony Blair Institute’s report on UK quantum strategy, which has the UK third globally in academic research in quantum and second highest in terms of venture capital investment. There are huge reasons to be very positive and excited about the UK’s quantum future, but “the country risks failing to convert its leadership in quantum research into commercial scale and strategic value”. Jonathan, what do you think, and what do we as parliamentarians need to do to push that?

Unfortunately, the report missed a couple of things and could have included a treatment of the strategy that had been published. A core element of the strategy is to create a number of quantum missions, moving from collaborative R&D and research, which have been very successful over the last decade, to more strategic objectives that we can get behind and make larger, more focused investments in. Even though we do not have the details—I will come to that in a minute—the intent behind the missions was well known, but there was no mention of them at all. There are plenty of things you could argue about, but they represent the best way forward for the UK in terms of bringing our activities together, moving to the next phase, re-establishing ourselves and continuing as one of the world’s leading quantum nations. There was nothing on that in the report, so that was somewhat disappointing, if I am honest. It also made recommendations about investing in this or that company, which again I could not agree with. It effectively amounted to taking a punt.

That is fascinating. Thank you. I really appreciate that. Would you say that the UK Government are doing well in terms of supporting the quantum strategy?

The strategy is good—I understand it came from the previous Administration—and they are looking to continue with it, but we must be five months overdue now in getting the details of the programme that should start on 1 April, which is disappointing. I am confident it is coming, but industry needs to see the details. With that and the delivery in place—which is another unknown—we can go out on the global stage again and make the case.

The clock is slightly against us, so I will come to Mr Cooper at this point.

In one of our roundtables, we spoke to an old-school investor who started out with £10 and built up a tremendous business. He told me he would not put a penny into AI. He said it is all a bubble that is not going to deliver. We will not solve that question here, but the Bank of England is worried about a bubble and suggests that a bubble might burst. What would that mean for AI infrastructure in this country? That leads on to quantum as well, but if I could start with you, Hugh, what do you think about the idea of a bubble? Should we be concerned?

None of this constitutes investment advice, but you are under an obligation to be candid with Parliament.

In some respects, it depends on your business model, where you sit in the stack, what services you provide, who buys them and their added value. I can only speak for Microsoft: we have a presence at various parts of the stack. Chips are very expensive and hard to find, and there is a long lead time on them, but there is huge value in other parts of the stack. The analysis is still emerging, but what we can already see are the productivity gains from AI not as a question-and-answer machine but as a fundamental way of shifting and changing workflow. When it is integrated into workflow, the economic value is very evident already. There is plenty of proof of that. If you build and sell services that dramatically impact on productivity, I would not recognise that as being part of something that is essentially a bubble.

There is bubble conversation; you read it in the press. I can only go to the data and the information that I see in the seat that I sit in. There is such a level of demand that it has been referred to as violent demand. We see an amazing amount of demand from our customers. So the decisions we make to invest are made based on—

What is that demand? When we were in Newcastle, we heard of a well-known company that went to Newcastle University and said, “Give us AI.” They had no clue what it was; they just felt it was something they had to have. Is the demand focused properly?

Let me give you some data points. When CoreWeave announced our Q2 results for June, we had an order book of $23 billion, which is orders we had yet to fulfil because that is the amount of demand we see. At Q3, we had an order book of $55 billion. Large and small organisations say, “We want to get access to this high-level computing which you guys can both construct and build and deliver as a service.” From that respect, the demand is definitely there. We build infrastructure based on either actual demand or demand that we can sell as soon as it comes in. To a question that was asked earlier, it is a challenge to make sure that people understand how to take that high-performance compute and implement it into their businesses. We made a recent acquisition of a business in the UK called Monolith, which has individuals who interact between AI, engineering and physics in the physical environment. They understand how to put AI into the hands of people who do things in the economy. The reason I mention that is because the question was asked, “Are we just selling these businesses and taking them to US companies?” We take that business in and then rather than it working on 10, 12, or 15 individual customers on a consulting basis, we can bring those people in and forward deploy them into a much wider number of customers who are trying to understand and use the capability. I look at that level of demand, what drives it, and how that can work into the wider economy.

I do not believe the valuations of AI companies will change the fundamental potential of quantum and quantum computing in particular to enhance AI. I worry about the spillover on to other deep tech such as quantum. Earlier we talked about the UK’s companies; if this started to impact investors’ views on the valuations of quantum tech companies and whether to invest, that would be an issue, but it would be a spillover.

That has been extremely helpful, thank you. One of the things I will take away from this conversation is that the fundamentals are pretty good when it comes to UK-US co-operation in this space. Although the technology prosperity deal and other agreements may have de-risked some investment decisions, the UK should not lament too much the suspension of the deal right now; we should focus on getting the policy environment as good as we can. Is that a fair summary of the views?

Yes.

Yes.

Yes.

We have beautiful clarity; I love it.

We will conclude with that. Thank you very much indeed for your evidence. That concludes today’s panel.