Jul 1, 2024 · 1h 6m · news
Matt Clifford: The Bull & Bear Case for China's Ability to Challenge the US' AI Capabilities | E1172 · 20VC with Harry Stebbings
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
In this extensive episode of 20VC, host Harry Stebbings interviews Matt Clifford, co-founder of Entrepreneur First, to analyze the limits of current AI scaling, the geopolitical tensions of the US-China tech race, strategies for spotting elite entrepreneurial talent, and the regulatory and infrastructural hurdles facing the United Kingdom.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Harry holds 15.5% of the talking time here. How this is scored →
Harry as informed peer 4.5 Guest teaching 4.7 Guest disagreement 2.3 Harry pushing back 3.0
speaking balance: gold is Harry, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 40:00 Flat refusal of AI-nuclear weapon equivalence
Matt forcefully rejects the premise popularized by Alex Wang that current AI capabilities exceed the destructive power of nuclear weapons, humorously asking Harry if he has ever tried a hydrogen bomb.
Hardest push from Harry ▶ 7:29 Challenging startup survival in OpenAI's pathHarry directly challenges Matt's investment thesis by arguing that startups attempting multimodality are placing themselves directly in the steamroller path of OpenAI's upcoming feature releases.
Biggest teaching moment ▶ 34:45 Revealing real-world UK infrastructure blockagesMatt educates Harry on the practical barriers to UK tech growth by sharing direct conversations with Satya Nadella and Andy Jassy, who reported local councils blocking multi-billion-dollar data centers over railway views.
Harry holds his own ▶ 14:38 Demonstrating venture expertise on valuation plateausHarry takes control of the conversation by laying out a sharp counter-thesis against OpenAI's $90 billion valuation, detailing how model switching behavior and GTM plateaus undermine the price tag.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Harry as informed peer | Guest teaching | Guest disagreement | Harry pushing back | Why |
|---|---|---|---|---|---|---|
| Formative Years: Matt's Bradford Childhood and Early Entrepreneurship | 3 | 2 | 1 | 1 | Harry opens with a personal prompt about childhood adversity, prompting Matt to share his background growing up in an ex-industrial northern English town. Matt describes fixing a damaged computer at age 13 to avoid working at Greggs, realizing he could accomplish things without seeking permission. | |
| Decoupling CVs: Predicting Founder Success Across Cultures | 4 | 5 | 2 | 3 | Harry asserts that top founders start early, asking if Matt disagrees. Matt agrees broadly but reframes the premise by explaining how cultural contexts alter founder signals, before transitioning to discuss tech S-curves and commoditization. | |
| Defeating Giants: How Startups Can Compete in AI with Ideas Over Capital | 5 | 5 | 3 | 4 | Harry pushes back on Matt's thesis by pointing out OpenAI's aggressive multimodal product updates and asking Matt to choose the primary AI bottleneck among compute, algorithms, and data. Matt elaborates on how video data and world models represent the next prospective S-curve. | |
| Diverging Capabilities: The Fallacy of Continuous AI Hype Cycles | 5 | 5 | 3 | 4 | Harry cites an essay by Alex Schultz on tech hype cycles and points to Chinese AI models as evidence of rapid commoditization. Matt argues that future model generations like Llama 4 or GPT-5 will require fundamental algorithmic divergence rather than simple compute scaling. | |
| OpenAI's $90 Billion Valuation and the Race to the Bottom | 6 | 6 | 3 | 5 | Harry asks Matt if he would buy OpenAI at a $90 billion valuation, offering his own detailed rationale for why model commoditization and plateauing GTM make it unappealing. Matt shares inner details from negotiating with senior Chinese government ministers, noting the CCP's extreme paranoia surrounding AI safety and domestic stability. | |
| Information Asymmetry and the Compound Effect of Chip Embargoes | 5 | 5 | 2 | 3 | Matt highlights information asymmetry between Chinese and Western tech ecosystems and explains how chip embargoes compound over time. Harry notes bipartisan continuity in US export controls across both Trump and Biden administrations. | |
| The Protocol Layer: Structuring the Economy of AI Agents | 5 | 5 | 2 | 4 | Harry cites Adept founder David regarding vertical integration across the AI stack. Matt uses high-frequency trading as an analogy to argue that protocol infrastructure for autonomous AI agents will become a massively valuable business category. | |
| Skills-Biased Technology and the Threat of Wealth Inequality | 5 | 5 | 3 | 3 | Harry raises concerns over technological skill gaps and growing wealth inequality in the UK. Matt criticizes the EU AI Act as overly restrictive legislation drafted by Brussels bureaucrats, contrasting it with the UK's light-touch regulatory regime. | |
| Overcoming Red Tape: UK's Infrastructure Bottlenecks and Pension Fund Allocations | 5 | 6 | 3 | 4 | Harry questions why the British tech ecosystem suffers from persistent negativity. Matt counters with optimistic economic projections while sharing anecdotes about US tech giants facing council vetoes on UK data centers over local scenic views, alongside a critique of zero-percent venture allocations in UK pension funds. | |
| AI Defense Tech and the Underrated Threat of Nuclear War | 5 | 6 | 4 | 4 | Harry brings up defense startup Helsing and asks if AI is more dangerous than nuclear weapons. Matt directly rejects the comparison to nuclear weapons, citing Annie Jacobson's research, before analyzing asymmetric drone warfare dynamics. | |
| Defending Western Values: Resisting Complacency and Degrowth | 5 | 5 | 3 | 4 | Matt strongly condemns economic degrowth and stresses the necessity of defensive technologies to protect Western democratic values. Harry pokes fun at venture capitalists using corporate consulting jargon like 'kill chains' without deep defense expertise. | |
| Co-Founder Dynamics and the Illusion of the Default Founder | 6 | 6 | 3 | 3 | Harry relays a question from Charlie Songhurst regarding founder synergies versus individual peak performance. Matt cites the 'Forced Entrepreneurs' study to demonstrate that recession-pushed founders often outperform voluntary founders, contrasting US founder culture with UK grads joining trading firms like Jane Street. | |
| Belief Capital and Talent Allocation: Europe vs. US | 5 | 5 | 3 | 4 | Harry challenges Matt's view on ecosystem differences by pointing out high-profile UK tech exits like Magic Pony. Matt explains that Silicon Valley benefits from higher 'belief capital', framing the career choice as playing zero-sum games versus building transformative companies. | |
| Assumptions and Curation: What EF Got Wrong About Talent | 4 | 5 | 1 | 2 | Harry asks what Entrepreneur First miscalculated in its early assumptions about talent. Matt admits EF mistakenly over-indexed on experienced 30-something candidates early on rather than backing exceptional 21-year-olds, explaining their multi-stage selection funnel. | |
| The Charlie Songhurst Approach to Early-Stage Investing | 4 | 4 | 1 | 2 | In quickfire format, Harry asks which venture investor Matt respects most, leading Matt to praise Charlie Songhurst. Matt breaks down Songhurst's ability to evaluate peak founder talent and pitch back an amplified, hyper-ambitious version of a founder's vision. | |
| Quickfire: Contrarian Advice and Changing Minds on AGI | 3 | 3 | 2 | 1 | Matt shares contrarian advice on co-founders, explains why he changed his mind about new startups building AGI companies, and offers candid reflections on the emotional discipline required in fatherhood. | |
| Immersive Murder Mystery Games: Scratching a Human Itch | 1 | 2 | 0 | 0 | Matt reveals his niche passion for writing immersive historical murder mystery games to satisfy human needs for unscripted agency. Harry jokes about the creator economy in a friendly wrap-up. |
Statements from this episode (46)
Matt Clifford: Value of Scaling Compute and Data for LLMs Is Flattening
“I think that we are seeing the flattening off of the value of just adding more compute and more data to language models.”
Matt Clifford: China Is More Paranoid About AI Safety Than Other Governments
“China is more paranoid about AI safety than probably any other government.”
Matt Clifford: US Export Controls Successfully Restrict Chinese GPU Clusters
“The U.S. Export controls on, you know, the semiconductor supply chain have had an impact. It's harder if you are a big Chinese company to build a access a 100,000 GPU cluster.”
Matt Clifford: Not Attempting a Startup at Stanford Is a Negative Signal
“If you grew up in the Bay Area, and you go to Stanford, and you graduate, and you never thought about starting a company, and there's a very negative signal on you as a founder.”
Matt Clifford: AI Progress Has Come From Compute and Data, Not Ideas
“Certainly the last five years, but arguably longer, the story of AI is really a story largely about the deployment of just enormous amounts of compute and enormous amounts of data, and not that many new ideas, candidly.”
Matt Clifford: Startups Should Not Attempt to Build Larger LLMs
“Frankly, I don't think startups today should be trying to compete on building bigger and bigger LLMs. It doesn't make sense. It's mainly a capital game.”
Matt Clifford: Text-Based AI Scaling Is Hitting an S-Curve Limit
“I don't think there's a lot further to go in just finding more text. You know, I think we may be at the flattening out of the S-curve on text, but I suspect the next S-curve could just involve finding ways to yeah, to use video, to use, like, interactive exper…”
Matt Clifford: Capital and Talent in AI Labs Will Unlock Next S-Curve
“My, if I had to bet, my bet is that the enormous amount of capital and talent That has been aggregated, you know, around this relatively small number of companies because of the hype will be enough to unlock the next S-curve, and so you'll see, you know, kind …”
Matt Clifford: Pure LLM Approach Will Be Quickly Commoditized
“I think the pure LLM approach absolutely is commoditized. Will be, either is or will be very quickly commoditized.”
Matt Clifford: Meta's Llama 4 Will Incorporate New Architectural Ideas Beyond Scaling
“Llama-IV will not just be You know, a bigger Lama three. It will have to be different. It will incorporate new ideas.”
Matt Clifford: AI Lab Capabilities Will Diverge Over the Next Two Years
“In the short term, I think we're likely to see more divergence, you know, over the next couple of years than convergence.”
Matt Clifford Would Not Invest in OpenAI at $90 Billion Valuation
“Ah, you know, I think it, I think I would not be, I mean, I'm a huge fan of OpenAI.”
Harry Stebbings: Top Companies Will Be Defined by Model Switchability
“And I think, you know, the best companies in the world will actually be defined by their ability to transition between models.”
Matt Clifford: China Requires AI Developers to Submit Training Data to Government
“For example, if you train a large model, you have to, like, supply random samples of the training data to the to the CCP or to the government. You have to you have to, like, show it's, like, political, that it is not gonna undermine any of the, you know, state…”
Matt Clifford: CCP Restrictions Hinder China From Producing Untrammeled AGI Founders
“And I think in China, as we've seen in other parts of the tech industry in China, the CCP is very nervous about entrepreneurs being given that much room. And so I think that it's not trivial to create the conditions in which a Chinese Sam Altman or a Chinese D…”
Chinese AI Developers Translate Western Research Faster Than the West Translates Theirs
“When Leopold Aschenbrenner published his situational awareness thing, which obviously, like, you know, became You know, kind of blew up a couple of weeks ago. Within three days, I hear from a friend who follows China very closely, there were five Chinese circu…”
Matt Clifford: Next US President Will Maintain Aggressive Nvidia Export Bans
“The one bipartisan issue in Washington Is screw China. And, you know, like, it's certainly true that I remember writing about in 20, you know, 1819, about, like, Trump export controls, and what's interesting is that the Biden White House has been as aggressive…”
Matt Clifford: Highly Capable AI Agents Will Emerge Within Five Years
“I think we'll get really good agents in the next five years.”
Matt Clifford: Infrastructure Protocols for AI Agents Will Create Massive Value
“Whoever builds the protocol, that for key slices of the economy, Actually allows agents to come together and do business together. That's going to be an extraordinarily valuable company.”
Matt Clifford: The EU AI Act Is a Mistake for AI Innovation
“I mean, I do think the EU AI Act is a mistake.”
Matt Clifford: UK Has Less AI Regulation Than Any Major Tech Nation
“On AI the UK has less regulation than any other country that has a significant AI industry.”
Matt Clifford: London Retains Top AI Research Talent Better Than San Francisco
“I think in, in London, it's much more plausible that you build a truly world-class research team and keep them together for long enough to really see the impact.”
Matt Clifford: The UK Can Return to Being the World's Richest Country
“I think the UK can go back to being pretty much the richest country in the world per capita. I really truly believe that.”
Matt Clifford: Local Councils Blocked US Tech Giants' UK Data Center Investments
“And the number one issue that these companies had, and I'm really not exaggerating, was we want to invest billions and billions of dollars to build compute infrastructure in the UK, and your local county councils in the middle of nowhere keep vetoing our data …”
Matt Clifford: Wayve Is the Top Private Embodied AI Company Globally
“You have probably the top embodied AI company, private company in the world in Wave.”
Matt Clifford: Entrepreneur First Raised More Capital in Michigan Than London
“I've raised more money in Ann Arbor, Michigan than I have in London, and I won't name the name of the pension fund, but, you know, one of my biggest investors is a sort of state pension fund in Michigan, and they've been amazing partners to us for a long time,…”
Matt Clifford: Helsing Is His Best Angel Investment to Date
“My, certainly my best angel investment so far.”
Matt Clifford: Current AI Is Not More Powerful Than Nuclear Weapons
“I don't think AI today is more powerful than nuclear weapons.”
Matt Clifford: Inexpensive AI Drones Inherently Favor Rogue Non-State Actors
“Being able to build cheap, Smart drones benefits rogue actors, you know, non-state actors a lot more than it benefits states.”
Matt Clifford: Degrowth Is the Worst Acceptable Idea
“So I think probably like the worst idea in the world that it's acceptable to hold in polite company is like degrowth. Like the idea that like, oh, you know, like the real problem in the world is like Too many people, too much stuff, too much consumption, we sh…”
Matt Clifford: Pivoted Focus at EF to Defense Tech Over Market Potential
“One of the reasons I've, Really pivot my time within EAF to working on defensive technologies is, I think there's an amazing commercial opportunity there, but I also think it's what the world needs. We need to build the technologies that mean that the negative…”
Matt Clifford: Founder Peak Performance Matters More Than Co-Founder Synergy
“Synergies do matter, but I think probably the best predictor is how good is the best person at their best.”
Matt Clifford: Academic Literature on Entrepreneurship Offers Zero Alpha
“I would say, in general, the academic literature on entrepreneurship is Not helpful for either entrepreneurs or VCs. Might be interesting, but it's not, there's no alpha in it.”
Matt Clifford: Cambridge CS Graduates' Top Dream Job Remains Jane Street Trader
“Even today in the UK, even with all the progress has been over the last decade in the ecosystem I would say the most aspirational job for a Cambridge computer science grad is to go work at Jane Street and be a trader.”
Matt Clifford: Europe and India Would Rival Silicon Valley With Top Talent
“Like, why does Silicon Valley work? Because the most ambitious and talented people in that area become founders. If the same thing happened, In the UK, in Europe, in India, we'd have very similar outcomes.”
Matt Clifford: 60% of Imperial CS Grads Entered Finance When EF Launched
“When I started visiting university campuses properly, when Alice and I started EF I think it was Imperial where 60% of computer science grads went into finance. That number is not that anymore.”
Matt Clifford: Quant Trading at Jane Street Is a Zero-Sum Game
“You could be an extremely highly paid crossword puzzle solver which is sort of how I think about, Jane Street playing a zero sum game, or you can, like, change people's lives.”
Matt Clifford: Incubators Should Pick Exceptional 21-Year-Olds Over Good 31-Year-Olds
“You got into the thing where it's very hard if you're new to what we do to compare the CV of like a thirty-one-year-old Who's actually good but not exceptional, and a 21 year old who's exceptional but has no experience, and choose the 21 year old. But you actu…”
Matt Clifford: Entrepreneur First Cuts Half of Paired Founding Teams Before Funding
“For every hundred people that join at that stage, you know, probably something like 80 of them will pair up with someone, and sort of Come up with an idea that they, that they're excited about. But we then do a cut where we say, like, do you really want to spe…”
Matt Clifford: 66% to 75% of Pre-Seed Entrepreneur First Startups Raise Seed
“We now put in 250 care at that stage, and I would say of those, typically, like, two thirds to three quarters would raise a seed round.”
Matt Clifford: Charlie Songhurst Deploys More Capital Yearly Than Most Seed Funds
“He's probably deploys more capital than most seed funds, I would say, every year.”
Matt Clifford: Pitching Back an Ambitious Vision Tests Founder Conviction and Drive
“The other thing he does better than anyone else I know is he pitches back to the founder the most ambitious version of the idea they've just pitched him, and I think he, he's just exceptionally good at that, and he learns so much from how the founder responds.”
Matt Clifford: European Founders Self-Censor Ambition Due to Conservative Local Investors
“Part of the reason it's hard to do investing here is that often by the time you meet founders, they've conditioned themselves to, like, reduce the ambition in the pitch, because they're just, you know, being so used to people being like, well, that's not very …”
Matt Clifford: VC Preference for Familiar Co-Founders Is Survivorship Bias
“You can start a company with a stranger. You know, I think a lot of the conventional wisdom on who you should start companies with is just survivorship bias.”
Matt Clifford: Ambitious Founders Should Consider Building AGI Companies Today
“I, 12 months ago, if I'd been on here, I would have told you, It's too late to build an AGI company, like that ship sailed, there's gonna be like three or four of them, they already exist. I now think the most ambitious founders today should seriously consider…”