UK regulators & LLM's - Can you see the problem?

The UK Competition and Merger Authority (CMA) realised a report this week on Large Language Model's ( which there refers to as Foundation Models) revealing their concerns that what they call an "immensely exciting" technology will be dominated by big tech and well positioned incumbents. https://www.gov.uk/government/collections/national-security-and-investment-act

The paper highlight 3 key interlinked risks to fair, open, and effective competition:

1. Firms controlling critical inputs for developing Foundation Models may restrict access to shield themselves from competition
2. Powerful incumbents could exploit their positions in consumer or business facing markets to distort choice in Foundation Model services and restrict competition in deployment
3. Partnerships involving key players could exacerbate existing positions of market power through the value chain

But does the CMA realise why powerful incumbents have so much power in the market is because building, training and operating Foundation Models is very, very expensive. The hardware infrastructure , the compute power, the data, the talent required to build and deploy Foundation Models is extortionate and so is the cost of developing non-trivial applications that deliver novel and innovative services on top of them. The only way this can be funded is through:

1. The deep pockets of powerful incumbents principally big tech - This article gives you a snapshot of how important is big tech's money to LLM powered startups - https://medium.com/b8125-spring2024/if-big-techs-are-all-buying-the-entire-llm-market-is-there-any-space-left-for-startups-66e16daec9ae
2. State and state actors such as Sovereign Wealth funds
3. VC and private investors

Leaving aside the fact that the UK has very few powerful incumbents with appetite to invest big in this wave of AI tech or the fact that our governments resources are unlikely to be directed into this space in a quantum that can make a big difference, we are therefore left with VC to pick up the challenge of funding FM and FM applications. But there is a problem, as the FT points out this week in this article https://www.ft.com/content/b7cd9329-c3fd-4408-8807-ce9c2939da6c that lack of exits means VC's are finding it hard to raise. And VC's are blaming "An inhospitable IPO market and regulators constraining M&A. Without exits, it becomes more difficult to persuade investors to part with their money". This is not just a UK problem ( although the data suggests the UK is being more effected than most) as the latest figures shows the first quarter of 2024 was lowest amount of money by VC's since 2014. - https://twitter.com/chrisharveyesq/status/1778871646578954464?s=48

So if the CMA are going to make it difficult for trade sales of UK AI startups or scale-ups to occur ( and the 2021 National Investment Security Act especially when applied to AI startups also does not help) you can start to see a problem. In fact the numbers of Mergers and Acquisitions across all sectors in the UK in 2023 declined by over 17%. So I wondered if the CMA see this reduction in investors and founders ability to liquidise positions and get some form of return through M&A as a success. And let's not go to the total lack of investment appetite to invest in IPO's or tech stocks in general on the London Stock Exchange or AIM.

In short, the consequence of the CMA clamping down on M&A activity especially in the AI space in the UK, is that they make it even harder for investors to get returns. So who is going to fund the development of Foundation Models or applications built on what the CMA keeps saying is an immensely exciting and impactful technology?

The answer seems to be big tech primarily from the USA and China or maybe "State Aided" national champions from the EU or Middle East. Is this what the CMA, the UK government and the UK public want? Is this the best way to ensure fair and open and effective competition in the UK?

I maybe bias but I think what would be a better question for the UK and its regulatory bodies is how can we increase investment into the building and deployment of Foundation models (and the applications they empower) in the UK (preferably those that are not owned, majority funded and/or controlled by overseas companies or foreign powers). More specifically how can the UK incentivise and support a massive influx of private sector investment into AI First startups so that we have a flourishing, diverse, impactful and highly competitive Foundation Model (and their applications) ecosystem.

I have ideas. I would be happy to share. I am sure others have as well. Is anyone in power in the UK and its government agencies willing to explore these ideas? Or are we doomed to trying to regulate other countries AI technology?