The United Kingdom has just made a risky move: redirecting pension funds toward AI startups. This massive experiment could redefine how innovation is financed in Europe. Or, put another way, it could serve as a historical warning about the risks of mixing patient capital with high-stakes bets.
Photo: Gary Butterfield on Unsplash
In March 2026, Liz Kendall, Secretary of State for Work and Pensions, announced the "Pensions Growth Compact." This scheme will allow up to £50 billion of British pension funds to be redirected toward venture capital, with a special focus on AI, semiconductors, and biotech. The bet is clear: if Israel has been able to get its pension funds to invest in technology without issues, why can't the UK do the same? However, there are crucial nuances that separate political rhetoric from actual execution.
The Problem No One Wants to Admit Out Loud
The British tech ecosystem faces a black hole in growth capital. According to data from Tech Nation, UK startups raise competitive seed rounds compared to their European counterparts. However, when it comes to Series B and C rounds, the money dries up or is funneled to American funds. As a result, British unicorns end up being British in name only.
British pension funds manage approximately £3 trillion. However, less than 5% is allocated to private equity or venture capital, compared to 15-20% in countries like Canada or Australia. Most of the capital is in bonds and public stocks, generating mediocre returns that barely keep up with inflation.
The Israeli Model Everyone Wants to Copy
Kendall has repeatedly cited the "Israeli model" as inspiration. Israel has mandated its pension funds to allocate a minimum percentage to local venture capital since the 1990s. Today, companies like Waze, Mobileye, and Check Point were born with patient capital from Israeli pension funds. Interestingly, there is a crucial difference: Israel has mandatory military service, creating dense networks of technical co-founders and a small market that forces global thinking from the outset.
In the UK, these structural advantages are not present. It has universities like Oxford, Cambridge, Imperial, and UCL that produce top-tier talent, but that talent has historically gone to the United States for better salaries or remained in consulting and banking for stability. Capital is just one piece of the puzzle.
How the Pensions Growth Compact Really Works
Photo: Dorin Seremet on Unsplash
Participation in the scheme is not mandatory. This is the first friction point. Pension funds can choose to participate, but there is no legal requirement. The government offers three incentives:
Partial loss guarantees: The British Treasury will guarantee up to 25% of losses on qualifying VC investments. If a fund loses £100 million in startups, the government will cover £25 million. This is risk asymmetry sponsored by taxpayers.
Tax relief on gains: Capital gains generated by investments under this scheme will have an effective rate of 10% compared to the standard 20%. This difference is significant when discussing exits in the hundreds of millions.
Priority co-investment with British Patient Capital: The state venture capital arm will co-invest pound-for-pound with pension funds in specific deals, effectively doubling the capital available for selected startups.
The Funds Already Onboard
Three major pension funds have already confirmed their participation: USS (Universities Superannuation Scheme with £89 billion under management), NEST (auto-enrollment fund with £30 billion), and the local government pension fund (£360 billion combined). However, the mechanics are indirect: they invest in specialized funds-of-funds that then select the VCs.
This additional layer of fees has historically destroyed returns. An average pensioner will end up paying fees to the fund-of-funds (1.5%), to the VC (2% + 20% carry), and administrative costs. If the startup doesn't generate returns of 3x or more, the pensioner loses out compared to having been in an index fund.
Why AI Specifically (and the Risks No One Mentions)
The timing of this initiative is not coincidental. The British government sees a window of 18-24 months to position the UK as the European hub for safe and regulated AI, before Amsterdam, Paris, or Berlin solidify their post-AI Act ecosystems.
Cambridge is home to ARM, which is now public and more valuable than Intel. DeepMind was born in London, although Google acquired it. There is top-notch academic infrastructure in machine learning at Imperial, Oxford, and Edinburgh. And, crucially, the UK has adopted a pro-innovation regulatory stance compared to the precautionary approach of the EU.
However, here’s the less obvious risk: AI startups have brutal economics. They require capital-intensive resources for compute, data, and top-tier talent that commands salaries of $500K+ in senior roles. The moats are questionable when foundational models are commoditized. Additionally, exits are concentrated in mergers and acquisitions by Big Tech, not in IPOs.
The Elephant in the Room: Concentration Risk
If £50 billion ends up chasing the same 200-300 "hot" AI startups in the UK, valuations will inflate artificially. We are already seeing Series A rounds of £30-40 million for teams of 15 people with no revenues. When that capital comes from pension funds with horizons of 30+ years, the pressure for quick exits decreases. However, it also means that errors in capital allocation will take decades to fully manifest.
A partner at a top-tier VC in London commented off-the-record: "It’s fantastic to have more capital available, but if that capital comes with implicit political mandates to 'invest in British technology no matter what,' we’ll end up financing expensive mediocrity."
What This Means for Founders and the Ecosystem
For early-stage AI founders: There is more capital available, but also increased competition for deals. VCs managing pension funds will face additional pressure for governance, reporting, and reputational risk mitigation. Expect longer due diligence and more complex structures.
For emerging VCs: It’s an opportunity to raise larger funds without relying 100% on family offices or endowments. However, you will need to professionalize operations, have a verifiable track record, and accept stricter regulatory oversight.
For the ecosystem: If this works, we could see the first real wave of British scale-ups that don’t need to relocate to Silicon Valley to access growth capital. But if it fails, we’ll have diluted the returns of millions of British pensions to subsidize mediocre exits.
The Uncomfortable Question About Returns
Historically, venture capital as an asset class returns about 15% annually for top-quartile funds. Median funds barely outperform public markets. Pension funds need consistent returns, not occasional home runs. The VC structure is inherently power-law: 90% of returns come from 10% of investments.
Can pension funds tolerate extreme volatility and J-curves of 7-10 years before seeing positive returns? The evidence from CalPERS and other pension funds that experimented with VC in the 2000s is… mixed.
The Experiment Everyone Will Be Watching
Kendall has promised annual transparency reports on the performance of investments under this scheme. This is unusual in venture capital, where returns are measured over decades. However, it is necessary when the pensions of teachers, nurses, and public workers are at stake.
There are concerning precedents: Woodford Equity Income in 2019, a British fund that collapsed after overexposing itself to illiquid startups. Or the Spanish pension funds that burned themselves on Bankia. The difference here is that there are government guarantees that socialize losses.
If in 2029-2030 we see significant exits from British AI startups financed under this scheme, the model will replicate in France, Germany, and even Spain. But if we observe massive write-downs and governance scandals, it will kill the European appetite for mixing pensions with venture capital for a generation.
The British bet is not just about AI or startups. It’s a question of whether Europe can create alternative capital structures that do not depend on American or Chinese LPs. It’s about whether patient capital can compete with impatient capital in a world where speed to market defines the winners.
Should your pension fund invest in your neighbor’s AI startup? This is the question that millions of Britons — and soon Europeans — will have to answer.