18
Examination of witnesses
Antony Berg, Eleanor Lightbody, Barney Hussey-Yeo and James Smith.
Q21 The Chair: Welcome, everybody. This is the Communications and Digital Committee. We are continuing our inquiry into scaling up in the AI and creative tech sectors. Today, we will focus specifically on AI and AI-first businesses. Those who have been following this inquiry will know that our first hearing, a couple of weeks ago, was very much a scene-setting hearing. We looked at a range of issues; a range of angles was represented by the various witnesses we saw that day.
Today, we have two panels. The first panel is very much made up of representatives from the coalface, if I can put it like that—founders or leaders of businesses that are either scale-ups or start-ups aiming to be scale-ups. We will ask them questions about their experience of innovating here in the UK and their access to capital. There will also be some questions on the impact of the copyright and text and data mining regime, and on what role the Government can play in incentivising these businesses to stay in the UK.
First, I ask each of you to introduce yourself and briefly say something about the business you are representing today.
Antony Berg: I am the chief financial officer at Speechmatics, which was founded by Tony Robinson in Cambridge in 2006. We are an AI speech technology business specialising in automatic speech recognition; we are market leaders there, with high accuracy across multiple languages and accents. More recently, we launched Flow, a conversational AI tool that is, I hope, leading on interactions between humans and technology.
Barney Hussey-Yeo: I am the founder and CEO of Cleo AI. Cleo is an AI assistant that helps normal people to manage their money. For context, we are a business with $150 million in ARR. We are profitable and growing to X year on year. We have raised around $200 million in capital. We operate 100% in the US. We were founded in the UK then moved to the US.
Eleanor Lightbody: I am the CEO of a company called Luminance, which was founded by mathematicians from the University of Cambridge in 2015. We have built a specialist AI that can help businesses with any interactions they have with their legal contracts. This means that customers—including AMD and, closer to home, DHL—use our AI on a daily basis to automate and augment the creation and negotiation of legal contracts, and to find things in their existing contractual landscape.
James Smith: I am the CEO and a co-founder of Human Native AI. We are a data marketplace that helps AI developers to get access to data in a fair, responsible way and enables rights holders to receive compensation for their works when they are used to train AI models. We were founded in the UK earlier this year; we raised seed money from UK-based venture capitalists.
Q22 The Chair: Thank you, and welcome again to all of you. I will kick us off with a bit of a high-level question. I am keen to understand where you think the main opportunity for AI innovation in the UK is and what the stakes are. I am particularly interested to know whether you think the UK should have a UK-based foundation model, to compete with the likes of OpenAI and Google, or whether you think there is scope for us to be more specialised in the products and services we develop on top of those foundation models.
Eleanor Lightbody: Let us take a step back and think about the UK. We have such a rich heritage when it comes to technology and artificial intelligence. I am biased, but we have some of the best ingredients to build transformative AI companies. Indeed, we have already done that.
When I think about foundation models versus specialist models, I am biased because we have built a specialist model for a very specific application, but the reach of what we are doing is horizontal. There is a huge opportunity to build these specialist models for what the UK is good at the moment, as well as for things that we do not necessarily know about. Is that not the whole excitement of AI—that it is transforming our lives in a way we cannot predict for the future? We should be investing time in that as well as investing in augmenting or automating some of the professions where we are so strong. For Luminance, that is the legal profession, financial institutions and professional services. There is so much that we can point our experience at and hire from very easily.
The Chair: Mr Berg, what is your view?
Antony Berg: The UK is relatively well placed. We have to remember that the tech sector alone is probably worth £1 billion in value today. That is more than in any other European country. We talk about investment in having our own LLM. We have heard about—James brought this out—DeepMind being a business that was originally founded in the UK. I think we are relatively well placed. We have strong universities and strong VC, particularly in the early stage, from investors. Where we need to get to is probably more on the growth and investment side, but we still attract huge amounts of investment from the US and other places. I still think that the UK is really well placed for AI investment.
The Chair: On a specialised route or also on potentially growing our own large foundation model?
Antony Berg: I think that going specialised will be the way forward. It is there already: you have the likes of OpenAI and all the others on the west coast that have developed these LLMs. It is about how you tune those and make them specialist for different sectors, whether that is legal or healthcare and so forth.
Q23 The Chair: Thank you. Mr Hussey-Yeo, what is your view?
Barney Hussey-Yeo: It is slightly different. I think we have missed the boat on foundational models. In the UK, we have no tech champions. We have no tech company listed on the LSE that is bigger than £10 billion in market cap. The US has five trillion-dollar companies—Nvidia is bigger than the whole LSE. The US gets to build these foundational models because it has the capex, the talent and the people who can invest in it. If we were in France, I would say, “Hey, you’ve got Mistral” or “Hey, you’ve got nuclear power. We could legitimately build these foundational models”. There is no chance that we will get nuclear power going in the UK or be able to build these companies now.
We have missed a huge part of value creation and value capture over the last 20 years, but the UK has incredible talent and universities. London is the second-best place in the world, after San Francisco, to build an AI company. In terms of talent density, there are more researchers here than in New York or anywhere else. We also have an amazing early-stage venture ecosystem. Over the last 10 years, we have built one of the most impressive venture ecosystems in the world and it is growing massively.
The challenge that we have as a country is that we cannot scale these companies, list these companies or capture their tech value, and if we do not do this the UK will continue to stagnate for decades to come.
The Chair: Thank you. Can you explain a little more what you mean by these VC eco-companies?
Barney Hussey-Yeo: They are venture capital ecosystems. For early-stage venture capital, London is the second-best place in the world to go and raise that. It is better than New York or anywhere else, outside of San Francisco. We have a huge amount of early-stage capital with great VCs. All the US VCs are ploughing in here. This is an amazing place to get a company going and start a business. It is a terrible place to scale and list a business, though. We have done amazing work at an early stage and now we need to move on to the scale-up stage.
The Chair: Thank you. Mr Smith, what is your view?
James Smith: I would echo some of the comments of my fellow panellists. Speaking as an early-stage founder, this has been my first attempt at starting a company. I used to work for Google and DeepMind, so I have some experience of working inside these very large companies. As a new entrepreneur, I have been very impressed by the community around raising money, the willingness of people to invest in the entrepreneurs in the UK, and the recognition of the talent that we have. I hope to be in the position of some of my fellow panellists in having challenges with scaling going forward, but some interesting questions get asked to us even now. I was in San Francisco a few weeks ago, and when VCs speak to us the simple question is: “When are you moving here?”.
Q24 The Chair: All right, we will come on to why you should not. Let us just focus. From what you have said, there seems to be a unanimous view across the panel that it is about the specialised way forward, rather than trying to focus on competing on a foundation model. Mr Hussey-Yeo made an interesting point about the lack of nuclear power here having quite a big impact on our ability to start up a foundation model.
I am interested to know what you think in headline terms. Tell me more about the barriers; I come back specifically to you, Mr Hussey-Yeo. Where do you see the barriers to realising the kind of potential that you have outlined to us?
Barney Hussey-Yeo: Take Revolut, for example. Revolut will be the first UK-founded $100 billion market cap company. It is fair to say that it will probably get to £1 trillion. Nick, the founder of Revolut, has said publicly that he will not list on the LSE; he will list on the NASDAQ. He has moved to Dubai, so he will not be paying any tax to the UK, and the vast majority of Revolut’s staff are outside the UK now. So even though that was a great British company, founded in the United Kingdom, regulation pushed him out of the UK. It was so hard for him to get a banking licence here—he got 20 others globally—that it forced him to move out. He took the opportunity of moving to Dubai and he hired people elsewhere, so we have missed an opportunity there. That could have been a great British success story—our first publicly listed tech company with a market cap of more than $100 billion—and the founder has left the country. He is not going to list his company here.
If you are looking for the next generation, there a lot of these companies that could now go public, by the way, but they are not going to go public in the UK because there is no capital going into the growth stage or into the LSE. Compare us to the US. In the US, pension funds are investing and there is a huge amount of capital going into the growth stage. In the UK, a growth fund is about £650 million. In the US, it is $8 billion. You can barely raise any growth capital in the UK. The amount of it here is anaemic, and you would never choose to list on the LSE because there is no depth of capital. The pension funds are really risk-off, so you will list on NASDAQ if you want to be one of the best companies in the world and built from the UK.
That is the sad fact we are seeing. ARM has gone to the United States and DeepMind was acquired by Google. All the value created by British companies will go to the United States and other foreign powers. We have to capture value and create more of it in the United Kingdom today.
The Chair: Thank you. We will come on in a second to more questions about capital. Do the other witnesses want to emphasise or highlight anything else as a barrier to realising potential?
Eleanor Lightbody: To add to Barney’s point, I see this through a different lens.
The Chair: Can you stay off it if is something other than this? We will come on to capital in more detail later.
Eleanor Lightbody: No, it is about the barriers. It is about celebrating growth here in the UK versus the US. In the US, they have a mindset of growth compared to here in the UK. My background is that I was at Darktrace when it was small and I saw it grow until just before the IPO. I obviously know Poppy very well. I also know Martina from Featurespace very well, and she has just sold her company to Visa for $1 billion—from what I read; I only saw that hit the headlines once nationally.
We need to do better in the UK to celebrate some of these successes and build people up, because if we do that we will inspire the next generation. The more we celebrate what is a hard thing to do, the more we will inspire and the more companies will be built. The more companies that are built, the more there will be an injection of capital and the more there is to celebrate. That virtuous cycle of success will get the UK further and further ahead, but without that celebration it just makes it even harder to do.
The Chair: Does anyone want to raise anything about compute or anything else?
James Smith: Access to compute is really a capital issue. It comes down to how much you can spend on compute and how much access you have. Another barrier to growth that I might raise is certain regulations, which I believe we will cover later, particularly on tax and data mining.
Antony Berg: Just on compute, the delta is huge. Look at the large companies: Google, Meta, Amazon and Microsoft are spending £250 billion a year.
The Chair: And do you have access to that? When you say that accessing the compute is about capital, is that because there is no compute power beyond those big companies?
James Smith: If we wanted to build a model and scale that model, the very first place we would start is looking to access one of the large cloud services run by some of the companies that my co-panellist mentioned. That is easy to access. Effectively, anybody can have access to it today. If you want to access huge amounts of that compute, you would look to do some more custom deals where you would have access to certain kinds of hardware. That hardware is available, although there are probably shortages of it at the moment, but it is generally an issue of having the capital to be able to access it.
The Chair: Finally, is there anything you want to add to what Mr Hussey-Yeo has already said about regulation as a barrier? Does anybody have any thoughts about pending regulation? Mr Hussey-Yeo, you look like you are bursting to tell us something.
Barney Hussey-Yeo: I have lots of thoughts on regulation. One of the amazing things about building a company in the US and trying to scale it there is that the checks and balances of the regulators there are incredibly effective. They have been incredibly thoughtful in how they constructed the country. In the UK, if the FCA says, “Hey, you can’t do a thing”, you are done—and done in the entire nation. In the US, you have the federal and the state, and even if you have problematic regulators in any position, there are these checks and balances.
That enables companies to scale and thrive, and if we still have these regulators with all the incredible power they have today and they are not risk-on, if they do not have a growth mindset and are not supporting innovative companies, the companies will move to Europe or the US, instead of building their companies here in the UK. We need regulators that have a growth mindset—with checks and balances. We need to co-operate with the United States and Europe to have global regulation, so that you are not having to understand the regulation in every geo as you scale your company.
Antony Berg: There is one more point. It is interesting regarding Silicon Valley. Bill Gurley, a partner at Benchmark Capital, does a presentation about Silicon Valley being 2,850 miles from Washington, DC. He says that they need to be as far away from regulation as possible in order to innovate, and that regulation can slow you down.
Q25 Lord Hall of Birkenhead: Just a quick question for Mr Hussey-Yeo. You are clear that we have missed the boat on compute. What would you have to believe to say that we had not missed the boat and you could do something about that? Obviously cheap power.
Barney Hussey-Yeo: You need an Apollo-style mission, a Manhattan project-esque mission, to be able to do this. Maybe that would have to happen. But if you are going to build these data centres, these foundation models in the UK, you first need power. We have the highest electricity cost across all western geographies. We would have to build nuclear. Renewables would not do it.
Secondly, you would need a Mistral-esque company, backed by venture capitalists and then supported by the UK Government, or you would build state capacity. You would take a state department and say that this is of national importance. This is one of the most important technologies of our lifetime. We should have a Manhattan-style project that spends billions and billions of pounds, a huge percentage of GDP, to compete globally with those big global tech giants. It is outlandish. Are we really going to get the Government behind a Manhattan project-style programme?
Lord Hall of Birkenhead: Is it outlandish and therefore forget it? Or, if you were thinking about the next 20 years—
Barney Hussey-Yeo: Of course you would do it. You would build nuclear today. You would build this project today. You would put billions and billions and billions towards it. You would hire the best people. You would build a venture-backed company with the right CEO. You would get at it today if you thought that this technology was going to be as transformative as we believe it will be. This would be such an obvious thing to do. Are the UK Government going to be able to build nuclear power plants in two or three years? Elon Musk built one of these AI data centres in 19 days. To build a nuclear power plant in the United Kingdom today would take you a decade or two decades. It is not going to happen. I think France could do this. I just do not think the UK could do it, given our Government and where we are at today.
Baroness Harding of Winscombe: I was going to double-click on the power question, which we have just covered.
Q26 The Lord Bishop of Leeds: Thank you. I hope we will come back to the question of what is unique about the UK mindset that hinders some of this. I should like to drill down into the question of accessing growth capital, because we are hearing mixed messages from companies that say that it has been okay accessing capital, and others who say it is impossible. Can you say something, starting with Mr Berg, about your experience of accessing growth capital. There is clearly a gap. You can see a great place to set up a business, and then they all go. Why?
Antony Berg: I can talk to my most recent experience. At Speechmatics we did a £35 million primary growth round about two years ago. When we went out it was a growth round, so we were looking for larger funds. We spoke to UK-based funds and US funds. The interest we got was mostly from the US. We also had UK funds that were interested, but we went with the US fund. It is important to say that our existing other investors in the earlier stage, when we talked about series A and even seed capital, were fantastic and followed alongside our most recent investor. Ultimately, in our business, two-thirds of our revenue comes from the US, and we wanted not only US money but a fund that could help us to build our US presence and grow. They had that infrastructure and mindset as well to help us deliver that.
The Lord Bishop of Leeds: On to Mr Smith, who has recently started a new business.
James Smith: I would be happy to share my experience. We are an early-stage company about seven months into our journey, so we have so far only been looking for seed capital as part of our experience, and we have found the environment to be really welcoming. We have managed to raise money from two UK-based VC firms with a track record of investing in and growing companies. One of our investors, LocalGlobe, is one of the success stories in European capital in getting companies to the unicorn status of $1 billion valuations. We have also heard the area around London described as New Palo Alto, the alternative place to start a company compared to Silicon Valley. So that initial stage is actually very good.
There are a lot of foreign investment funds now here. They have a presence and are open to meeting new entrepreneurs from the UK. We have not yet reached the stage where we are looking for the next round of capital, but if we were at that point, one of the things we would be considering would be capital from outside of the UK, not necessarily because we do not have access to it but because it would give us the opportunity to connect with new networks and grow our business globally. So there are also positive things about UK companies receiving foreign investment at that stage.
Q27 The Lord Bishop of Leeds: Miss Lightbody, are there implications further down the line for this dependence on foreign scale-up funding?
Eleanor Lightbody: Our early investors were UK-based as well, and in our series B earlier this year we raised $40 million. What was interesting was that, this time last year, while we were raising, we had five offers on the table, four of which were US VCs and one UK. That probably represents the make-up.
To echo what we have heard, we were interested in US VCs for the network they had, the ecosystems they built, the people they had and the companies they could introduce us to—frankly, in their experience in taking companies from our size to billion dollar-plus. For us, at the moment, we feel strongly that that has worked very well, and that our competitive advantage is keeping our tech and our R&D team in Cambridge. We hire directly out of the University of Cambridge and feel that that is one of our competitive advantages. Thinking about growth, the US is of course the most important market for us, but we feel super-strongly about keeping our technical know-how here as well.
The Lord Bishop of Leeds: Let me hear from Mr Hussey-Yeo before I come back on that one.
Barney Hussey-Yeo: My quick take is that early-stage venture works incredibly well and is well capitalised in London. It is an incredible ecosystem. It is performing. The problem is that we have no LPs in the UK that will invest in venture funds. So the pension funds do not invest. There is no capital going into the growth stage. The growth-stage funds in the United Kingdom are extremely tiny. If I want to raise £200 million-plus, which will be my next round, there is no fund in the UK that can support that level of capital. We have to go to the United States to raise that capital, because pension funds do not invest in venture. There are LPs in the UK that are not investing in an amazing asset class, and that is why our pension market is so poorly returning. I welcome the reforms that have happened with Rachel Reeves on consolidating the pension funds, but we need growth capital to scale these companies or they will all move to America.
The Lord Bishop of Leeds: Is the hinderance to do with mindset or regulation, or what?
Barney Hussey-Yeo: The pension funds have no incentive to put money into venture. They want to make as much in fees and take as little risk as possible. The pension funds are sitting there having a great time making money, whereas they are screwing—sorry for my bad language—taking advantage of the UK population. The returns of the UK pension industry are diabolical. If you compare it to Canada, Australia or the US, they are investing in equities, in venture-backed companies. All the growth of the S&P 500 has come from technology companies. If we cannot get a growth mindset and risk-on as a nation, if we cannot get pension funds investing into venture, we will have another decade or two, potentially more, of stagnation. The UK has stagnated for so long because we have not been able to build, list and scale these tech champions, which the United States has.
Q28 The Lord Bishop of Leeds: I will just go back to the question I was going to ask, and then we will come back to that. You said you had four businesses offered cash from the United States and one from the UK, but is that simply a reflection of scale, because the United States’s economy, and the country, is so much bigger, making four to one reasonable?
Eleanor Lightbody: It is scale but also growth mindset and tolerance of risk. I am always amazed that here in the UK we get really excited by a $1 billion valuation—that would be one of the biggest tech companies to be built—whereas in the US that is just finishing chapter 1. If we can bring more of that energy here, that would be super-beneficial.
Q29 The Lord Bishop of Leeds: I hear what you say about pension funds and so on. What assessment do you make of recent initiatives to unlock more domestic capital here? Is the British Business Bank making any contribution to this? How would you evaluate that?
Barney Hussey-Yeo: My slightly contrarian view is that it is all fraud. If you look at the returns and where the British Business Bank invests its money—I note the Future Fund—every time the Government try to do direct intervention into venture, it does not work. Venture capital is about picking the outlier power law companies, so when the Government try to fund these companies directly, they pick the bad companies and, therefore, the returns are awful. The returns for the British Business Bank are negative to dismal, and that is £12 billion of capital that could have gone into fixing the NHS, investing in infrastructure or building supercomputers—you name it. Direct intervention by the Government into venture is a terrible idea. It is fraud, and we must stop it.
Antony Berg: I do not know enough of the detail, but, with the recent partnership, they are trying to help to release the pension funds to allow them to invest in the growth capital stage. That is what they are trying to do now. I think the intention is sometimes right. The Future Fund had the right intention: that was where they matched the VCs or investors during Covid to come up with a plan to keep that sector alive. The principle was right, although I do not know how good the execution was. They were doing it back to back with the VCs: they matched whatever the VCs were investing. So they were relying on the VCs making good investments. You could argue that the VCs were not making good investments and, therefore, the British Business Bank suffered as well. But the principle was right.
Eleanor Lightbody: At the time—a couple of years ago—we were probably exactly what the Future Fund wanted to invest in, but because of our shareholding make-up at the time we were not eligible for it. So that was quite an interesting thing to go through. In the last two years, our growth has been 5x, so that would have been a great investment, I think. But because of how we were set up, it was almost like, “Computer says no”. If we are going to do these things, we always have to have some agility, flexibility and, as you said, tolerance for risk, otherwise it will not succeed.
Q30 The Lord Bishop of Leeds: Can you briefly speculate on something? You have made clear that companies that have an ambition to scale up will not list in London. What would have to change to make London an attractive investment place to list your company in?
Barney Hussey-Yeo: If you are a founder, you are thinking about two things primarily. The first and most important is how successful your business will be. If I am thinking about listing and I talk to Goldman Sachs and JP Morgan, they will give me a big lecture on the depth of capital in the Nasdaq, why there is so much trading volume, why it is such a better place to list, et cetera. In the UK, we must get more capital into the LSE and into any tech company that lists there. But it is a bit of a chicken and egg; there are not many tech companies to put the capital in today.
The second thing is the personal gain of the founder—this is probably the most controversial thing to say. They are thinking about their tax position and what they stand to gain from this. The first thing is to make the company as successful as possible, and the second is: what is the incentive for the founder to list and stay in the United Kingdom? We just increased capital gains for founders, which is crazy. It is now better to go to Austin, Texas, and list on the Nasdaq than it is to stay in London. There is no incentive for founders to stay in the United Kingdom and list their companies here. So you have to get way more capital into the LSE, and that is a hard problem.
Secondly, you have to build a competitive system-level environment for founders to want to be here. Otherwise, they will go to Dubai or the US. Patriotism alone will not keep people in this country. We have to build the systems and the structure and get the capital into the LSE to keep people here.
The Lord Bishop of Leeds: There is another argument here that we will have to have another time.
Antony Berg: The IPO market is a route to exit for founders and to generate liquidity. We have to remember that the US market is not in a great place. Last year, there were 154 IPOs in the US, but in 2021 it was in the thousands. In the UK, there were 23 on the LSE last year, so there is a really small number of IPOs. The appetite for them is currently not there. I hope it will come back, but that is where we are today.
Eleanor Lightbody: We also need to see what happens when tech companies choose to list in the UK. Darktrace is a fantastic example: it just got bought for double the amount that it was at on the London Stock Exchange. The No. 1 duty as a CEO or a founder is to grow the business and return money to the shareholders, so if that kind of offer comes about, it will be very hard to say no to. It is one thing trying to incentivise people to list, but if they do, how do you try to keep them on?
James Smith: We at Human Native are still very early in our journey, so I hope these are problems we will figure out at some point.
Q31 Baroness Harding of Winscombe: I will push a little more on your arguments about how we compete by just being the same as the US. The US economy is 10 times our GDP and, given the scale of the customer and investor bases, we will never be the same size as the US. I want to understand how you would compare UK growth capital with other countries of a similar size, rather than just hoping that we could maybe be like the US, when it will always have a larger market.
Barney Hussey-Yeo: Europe has the exact same problems. It is having the same debate about scaling companies and trying to make them public. But the reality is that there are no growth funds big enough in the UK, because pension funds are not investing and are not building growth funds. If there were growth funds like those early-stage venture funds, UK founders would take capital from growth funds in the UK.
The biggest criminal thing that is happening is with all the LPs into all the British venture funds. The Canadian pension fund’s endowment is in the US. With our universities and our pension funds, normal individuals are not in equity markets like they are in the United States. We have to change this country’s risk mindset. Pension funds and normal people have to get investing. The US dominates, because they get all that capital into their public markets.
So if you built growth funds here, you would see founders raising from their growth funds. If you built an environment where it was more competitive as a founder to list on the LSE than Nasdaq, it would work. It is not all capital: you have to give a little bit of value capture to gain value capture over time. If you made it economically valuable for the founder of a UK-listed company to stay in the UK, they will obviously do so. In the United States, you will spend 38% on capital gains if you are in New York or San Francisco. That is a huge amount, which is why British founders do not all move to the United States today. I was very vocal when we were closing that gap, moving from 20% to 38%, because it would have killed the venture ecosystem. It would have moved all these founders out there.
We have the ability to change what is going on at a system level. We have the ability to make it more attractive for founders to stay here, and we have a huge amount of capital sitting on the sidelines. We have £300 billion in cash ISAs: the Government are incentivising people to keep cash savings instead of putting it into equities and high-growth funds. It is crazy and criminal what we are doing with the capital sitting on the sidelines that is not going into growth. If we changed that, it would make a massive difference to people staying in the UK, scaling companies and listing.
Q32 Baroness Healy of Primrose Hill: My question is for James Smith. Is the current lack of clarity on copyright and text and data mining a barrier to AI growth in the UK?
James Smith: I certainly think it is. There are some uncertainties in the current law. It is unclear whether a lot of the new models in generative AI infringe copyright under the current guidelines. I will leave the technicalities of that to the lawyers, but there are issues such as whether, for example, text and data-mining guidelines cover commercial or non-commercial research. There are other jurisdictions globally that have more clarity on those requirements, which is causing barriers.
We have heard about high-profile cases with copyright infringement lawsuits, where UK-based AI companies have admitted to training models outside of the UK, partly due to that uncertainty in the law. We have also heard privately from large technology companies that they are debating their investment in the UK, particularly in data centres, due to the lack of clarity in the law. So if the present Government were to do something positive in this way, my suggestion is that they should move to clear up that situation as quickly as possible.
Baroness Healy of Primrose Hill: Do you have any idea how they can do this incredibly difficult thing? They would have to navigate the trade-off between protecting the creative industries and still promoting AI innovation. The previous Government did not succeed with that, and although the present Government are still trying, they have not come up with a solution. Is there anything that you would say to them?
James Smith: Absolutely. It is a very tricky issue, and we should acknowledge that. Our company sits at the intersection of this debate, so we have been very fortunate to speak to people on both sides. We have had many conversations with AI companies about their sheer demand for data. A lot of the existing architecture around models is powered by getting access to huge amounts of information. They are looking to compete and to build the best possible products, so they need more and more data. We think that will continue to grow.
Equally, on the other side, the creators and rights holders feel angry about the current situation. There is strength in the feeling that their work may have been stolen. We talk about it as feeling like the Napster era of generative AI—for those of us who are old enough to have come across Napster. We need to think about what the new, more legitimate era of generative AI will look like.
On what the Government can do, I think that there is recognition on both sides, at least privately, that some degree of compromise is required. I do not believe that the public statements—for example, from the creative community saying that it absolutely will not support any future extension of the text and data-mining exception—will be borne out. We will likely move towards. or at least the Government have indicated that they may move towards, an EU-style text and data mining exception with an opt-out, which is a compromise in itself.
I believe that the AI community will probably have to accept a compromise on the transparency required to report what information they are using to train these models. If we can have some degree of compromise on both sides, we may be able to move forward towards an implementation phase, where many details will need to be worked out.
Baroness Healy of Primrose Hill: You think that an opt-out might be the way forward. Can you explain a little about how your business helps companies? It is not through a licensing agreement, is it?
James Smith: Yes, it is through a licensing agreement. We operate an AI data marketplace, which means that we help AI companies to get access to the data they need in a fair and responsible way. There have been some accusations of bad behaviour from the companies in the sector—for example, by scraping the internet or getting access to information where consent has not been granted by the rights holder. We think there is a large demand for access to data that is not available on the public web, or that is created or protected in some other way, and that creates an opportunity for licensing.
We have seen an interesting amount of demand on both sides of our marketplace. We have AI companies with a data acquisition problem, and which need to acquire large amounts of data quickly and at a fair price. On the other side, we have seen rights holders wanting to participate in this new AI economy.
I spoke recently to some media executives who said that they are at the end of the traffic era for their businesses. A lot of online businesses—for example, the newspapers in this country—have built their online businesses around receiving traffic from social media or search engines. That game is changing. A lot of the new platforms emerging are keeping users within their platforms and sending less traffic to these websites, which means that their traditional revenue streams of advertising and subscriptions are under threat. Therefore, perhaps a new economic model is needed.
Baroness Healy of Primrose Hill: Thank you. Would any other witness like to comment on copyright? No?
The Chair: I was waiting for someone to burst into an argument as to why this is terrible, but it is good that you are not doing so.
Q33 Lord Knight of Weymouth: James, I wonder whether you can help me to understand something. When we were in California—I think with Reddit—there was a question about whether AI can unlearn what it has learned. At another visit, there was a sense that when a new model is being trained, they train it with a data refresh. Is there anything that would lead you towards licensing a model so that it becomes a limited thing? Then you can renew your revenue source with licensing for the next model when that comes around.
James Smith: We are seeing a lot of innovation in the current licensing space for these types of questions. There are multiple different ways in which you could license content today as an AI developer. You could license content in perpetuity, which gives you the rights to use that content for ever if you like, or for however long you like. Those deals are relatively rare, and they are probably ill advised.
Lord Knight of Weymouth: An actor friend of mine had to sign one of those with Disney.
James Smith: Really? I do not feel that that is favourable to the rights holder in that situation.
We have also seen deals that are time-limited, usually to three or three years, which gives an AI developer the opportunity to train any number of models in that time period. The final way to license is, as you suggested, people licensing content for the lifetime of a model. We think there is a lot of opportunity there for more innovation and for things to change.
On unlearning data, various techniques are being explored, but I do not feel that that is a promising angle. It is better to start by building models from scratch. When there is a new generation of models, it is an appropriate time to reconsider the data set that you are building a model on.
I was in San Francisco a few weeks ago, where I had dinner with an engineer from one of the prominent foundation model companies. He said that, if his company were in a situation—through regulation or something like that—where it would be forced to throw away a model because it had used data that it was not permitted to use, that could be $1 billion mistake. So we think that many more AI companies will face increasing scrutiny from regulators on the information that is used to power their models, and that therefore they will likely have to prove that they have obtained that data legitimately and have the rights holder’s permission to use that model for that specific purpose.
Lord Knight of Weymouth: That is very interesting.
The Chair: We will move on to discuss incentivising businesses to expand while remaining in the UK.
Q34 Baroness Wheatcroft: You have all been very eloquent about the problems of scaling up. Honestly, businesses in every sector have problems with scaling up, but you at least seem to have managed to overcome them by getting the money from the States. Part of the message that has come through is that, as others have commented, even if the money were on offer here, it would have been very tempting to take it from the States, because you are looking at much bigger markets.
The next stage must be to ask how on earth we keep you based here, so that at least there are benefits flowing into the UK coffers. I would be interested to know what you think the Government could do, if anything. Mr Hussey-Yeo, I heard what you said about banking regulation, which is a very specific example, but is there anything on the regulatory front that you think the Government could do that would help AI companies to scale up generally?
Barney Hussey-Yeo: I have a simple suggestion. The incentive of regulators is not to take risk. They have no desire or need to take risk. If you have 100% authority and control, and you have no incentive to take risk, you will kill all the innovation. So there needs to be an incentive structure for regulators to take risk.
Baroness Wheatcroft: Which regulators?
Barney Hussey-Yeo: I am thinking very much of the FCA. I work in financial services. The FCA is the biggest regulator for the biggest market, et cetera, and if it had an incentive or a reason to risk, that would be a huge thing.
Secondly, there should be some sort of checks and balances on the regulator. There should be some discourse when the FCA makes a decision. There should be checks and balances here. Today, it has a huge amount of authority and power in the United Kingdom, as do the Government. That curtails lots of innovation. You need debate and the space to innovate and try new products.
Baroness Wheatcroft: Of course, companies that are not particularly involved in high tech have their own gripes with the FCA as well. Would you like to see any other regulators change their outlook a bit?
Antony Berg: There is also the Competition and Markets Authority. We had a position where Microsoft was trying to acquire Activision Blizzard. That took for ever. The acquisition ended up getting done, I think, but there was a huge amount of complexity there.
Now, when you want to get foreign investments, you have to go through the process of getting approval and check whether it meets certain criteria. The criteria are so broad you end up asking your lawyers, “Do you think we need approval?” They say, “Yes, let’s be conservative and get approval”. It just slows things down.
Q35 Baroness Wheatcroft: In your experience, is it tougher here than in other regimes?
Antony Berg: The US also has it tough. Recently, in some acquisitions, it has basically done quasi acqui-hire with licence arrangements in order to get around the regulations. The new Government there are planning to change that, I think, but that has been a big criticism there. The US has also had its challenges.
On the finance side, for me, it is also about the R&D tax regime. When you employ tech staff here, there is an opportunity for you to do R&D reclaim. There is a real tax benefit there, but they changed that regime. They said to the smaller companies, “You can’t have as much as the larger companies”. Previously, you could get back a third of whatever you were spending in R&D. That went down to 17%, and then they said, “Actually, that’s not great. There’s huge uproar. We’ll make it 23%”. However, larger healthcare companies could still do the reclaim, which is not okay. We need a stable regime. We need to get back to the R&D regime that we had before, because it worked.
The other part of this is that there is so much complexity in the criteria for that regime. There is, I believe, an easier way to make that process smoother.
Baroness Wheatcroft: Thank you. Ms Lightbody, is there anything that you can cite?
Eleanor Lightbody: I would echo the two points that we have just heard, but two other things I would mention are agility and making sure that you have a share of voice in the room when you are talking about regulations so that they are not being made in a vacuum or in isolation. Often, it will be the Government along with some of the Big Four or whoever, but some of the ones you might want to include are the two-man start-ups coming from the University of Cambridge and the University of Oxford, because they are the future of building.
Baroness Wheatcroft: Mr Smith, do you have anything to add?
James Smith: The point about R&D tax credits is an interesting one. As a company that is not yet profitable, those R&D tax credits are useless to us, in effect. Government financial support for early-stage, venture-backed start-ups that do not expect to be profitable for a number of years could do with some support.
Another example would be the Innovate UK grant system. We have been told that the maximum amount of grant we are eligible to apply for is £100,000. In many people’s money, £100,000 is a lot. Unfortunately, when you are building an AI company, that is about one engineers salary for one year. It does not help us to grow or scale in any particular way. We have been told that larger grants are available, but they are mainly for things in healthcare or other such aspects. If there were more opportunities in that space, you might see some more support for smaller, earlier-stage companies.
Baroness Wheatcroft: But without citing, “It’s £22 billion, blah blah blah”, there is obviously not that much money to hand over. Do you think there is any mileage in the Government being more generous with procurement contracts in the direction of home-grown AI companies?
Antony Berg: I am sure that we would all take it.
James Smith: One of the things I have heard—not personally but from friends in the business—is that there is a lot of difficulty in obtaining those contracts and a lot of time spent on negotiating them. If you have a small UK-based start-up with a few people from the University of Oxford, would you rather have them spend their time on products and developing their technology or on legal negotiations? If it were easier for these companies to be involved in a sandbox-type environment in order to work with some of the government contracts, I think that would be very much welcomed by the sector.
Antony Berg: Procuring with government is definitely a challenge because of—going to your point—the amount of legal work you need to do. Also, if you are a growth company and you have an alternative option in one of the mainstay larger companies, there is no incentive to take a risk with the smaller companies unless there is a big differentiation in quality of product.
Eleanor Lightbody: It might be easier to have more meet-ups, collabs and introductions. The Government speak to so many high-powered business owners on a daily basis. Getting some of the companies that are in scale-up mode or start-up mode together in a room would be somewhere to start. That could be super-beneficial.
Q36 Baroness Wheatcroft: Leading on from that, could the Government do more with smaller companies, such as taking them on trade missions and introducing them overseas? The Foreign Office used to be pretty good at that. Has anyone reached out in your direction saying, “Join this trade mission”?
Eleanor Lightbody: At Luminance, we participated in a few trade missions early on. They have a huge amount of impact when you are getting started. There is a bit of a gap, when you have so many customers, between now and what comes next—introducing you to that next layer—but, also, there are so many introductions that can happen here in the UK. There are so many business owners we can meet here. Whenever I go to the US, I am reminded of that. You go to these big events where everyone is opening doors for everyone else. We could do a lot more of that here in London.
Baroness Wheatcroft: Who do you think should convene that?
Eleanor Lightbody: It would be great if the Government could do some of it.
Baroness Wheatcroft: Is there anyone from the industry who might push this?
Eleanor Lightbody: I am sure that lots of people would. A lot of the people I speak to on a daily basis from the tech side of the world want to encourage more mentorship here in the UK. It is something that the US does really well. Founders or CEOs sell their company for $1 billion. They give back to the ecosystem. We see less of that here in the UK. There are fantastic examples of that happening, of course, but we should encourage it more. This goes back to the point about us inspiring people to build. It is a very hard thing to do. Let us try to remove some of the barriers that are there, at least.
Q37 Baroness Wheatcroft: The point has been made that keeping companies here is making it easier for them to bring in talent from overseas. You have all made it clear that there is plenty of talent in the UK. Do you think there is a need for an expedited visa regime so that you could bring in more talent if you were to stay here?
Eleanor Lightbody: Absolutely. A lot of the talent that you see in universities is not necessarily British passports. I can speak only for myself but, when I talk about talent, I am not talking about UK citizens; I am talking about those people who choose to come and be educated here. I know we have experience of trying to hire out of some of these universities. The paper trail and the process are so laborious. If you could remove some of that and remove the costs of doing it, that would be great.
Antony Berg: There could be some breaks around the costs of sponsorship. Eleanor makes a really good point. There are a lot of people coming out of university who have chosen to be educated here and need sponsorship to be employed here. The costs associated with that sponsorship and managing it are high.
Baroness Wheatcroft: Mr Hussey-Yeo, have you had the same experience? Not yet, perhaps.
Barney Hussey-Yeo: No. We hire lots of international people into the business. Post Brexit, though, that has declined. We have the second-biggest AI density in talent because of our universities but, since Brexit, fewer skilled migrants have been coming to the United Kingdom. It has become harder to get people into United Kingdom. People are choosing Paris and Berlin over London. That is not a good trend. If we cannot buck that trend and get people to want to come back to London and work here, France will be the dominant AI country in Europe and we will play second fiddle, which we are doing a bit today.
Baroness Wheatcroft: Thank you. On that note, I will come to a halt.
Q38 The Chair: This is a bit cheeky, but I want to ask Mr Hussey-Yeo a direct question. You talked at the beginning about your business’s trading in the US. What makes you stay in the UK? It is great that you are. I want you to stay.
You are not the first founder some of us on the committee have met in recent weeks, because some of us had the great fortune to go to San Francisco to hear how we are quite poor in this country at celebrating the success of our own founders. I ask you this question from the perspective of being very proud of the success of anybody who has started their business and scaled up, and I want you to say.
Barney Hussey-Yeo: If I am really honest, which I think I have been on this panel, one reason is the incredible amount of talent here. We have talent, and because of the stagnation of wages it is so cheap to hire amazing researchers from Cambridge, Oxford—you name it. They are three times cheaper than someone hired in San Francisco. I am here for the incredible, cheap talent that we have in the UK. That is a great thing, and we have a big advantage there.
Secondly, very candidly, the UK has a better tax regime than the United States does. We have 24% capital gains now versus 38% in the US, so I spend four months of the year in the US not to hit capital gains in the US.
There was one little point that I wanted to pull up on. To say, “The US is 10 times bigger as an economy. It has 10 times more capital, and we’ll never be able to compete with it”, is kind of missing the point of how venture works as a business. Venture is very much a neighbourhood, a local, business, so if you have those local venture funds, growth funds, in the United Kingdom, people take them. You build a relationship with the people over time, you trust them, and then they pre-empt you into your round.
So just because the US has more capital does not mean that if there was growth capital in the United Kingdom, founders would not take it. It is really important that we get pension funds and capital going into their growth surge.
Baroness Wheatcroft: Sure.
The Chair: Thank you. I am very grateful to all four of you for your candid answers to all our questions today. It has been fantastic to have you here. I will pause the transmission there, because we are going to swap panels and have a different set of witnesses next.
