Bill Maris: How Google Could Crush AI Competitors, Why Small Funds Win, and AI's Atari Stage
Bill Maris, founder of Google Ventures and Section 32, shares four lessons from his career spanning data center startups to venture capital, arguing that small funds outperform large ones. He discusses AI's current 'Atari stage,' Google's potential to crush competitors like OpenAI by slashing token prices, and the problematic incentive structures in modern venture capital.
Summary
Bill Maris opens with his origin story from 1997, when as a neuroscience graduate working on Wall Street, he discovered a server in an office closet and glimpsed the future of the internet. He quit his job and founded a web hosting company from his Vermont apartment, running servers in freezing conditions and famously tarring his roof during a thunderstorm to protect his equipment — illustrating his first lesson: sometimes entrepreneurship requires a degree of apparent insanity.
His second lesson comes from observing inauguration crowd photos across decades, noting how one person in 2009 was live-streaming on a laptop when no one else was — representing the kind of entrepreneur who 'knows a secret about the future that most of us don't believe.'
At Google from 2007, Maris built Google Ventures by collecting massive datasets and applying machine learning (Google wouldn't allow him to say 'AI' at the time) to design optimal portfolio construction and fund sizing. GV returned an estimated 4.1x overall, with his personally led investments performing even better — leading to his third lesson: don't bet against computer science applied to the right problem at the right time.
His fourth and most data-driven lesson is that small funds outperform large ones. Funds under $750 million averaged 4.76x DPI returns versus 2.42x for funds over $1 billion, and represented 95% of top decile performers. He illustrates the math: a $7 billion fund needs $210 billion in exits to return 3x, which exceeds total venture-backed M&A and IPO exit value in most years.
On AI, Maris argues we are at the 'Atari command line stage' and will reach a PlayStation-level equivalent within 5 years. He draws a parallel between how gaming evolved through better controllers, physics engines, and GPUs — not just better stories — and argues AI will similarly be transformed by infrastructure platforms, not just larger models. He is particularly interested in investing in these enabling layers.
Maris warns that Google could devastate OpenAI and Anthropic by cutting token prices by 80%, making their products commoditized. He is sharply critical of companies that stay private longer, keeping value creation among elite investors while claiming to benefit humanity, effectively making public 401k holders the 'bag holders' when these overpriced companies finally IPO.
On life sciences, he remains interested in computational biology but warns that human clinical trials and FDA processes mean biotech won't go as exponential as hoped unless realistic cell simulation in silico is achieved. He also expresses concern about US scientific brain drain due to anti-science policy sentiment and H1B visa restrictions.
Key Insights
- Maris argues that funds under $750 million averaged 4.76x DPI returns versus 2.42x for funds over $1 billion, with sub-$750M funds representing 95% of top decile performers — making small fund outperformance a mathematical reality, not an opinion.
- Maris claims Google could existentially threaten OpenAI and Anthropic by cutting token prices to 20 cents on the dollar, arguing that if an identical product is available at 80% less cost, competitor business models 'go super critical' under compression.
- Maris contends that AI is currently at the 'Atari command line stage,' and that just as gaming evolved through controllers, physics engines, and GPUs rather than better stories, AI's transformation will come from infrastructure platforms — not larger models — over the next 5 years.
- Maris criticizes late-stage companies staying private longer, arguing it forces overpriced assets onto passive ETFs and 401k holders who didn't participate in early value creation, while the companies simultaneously claim to be acting for the benefit of humanity.
- When building Google Ventures, Google prohibited Maris from using the term 'AI,' insisting it was 'science fiction a hundred years away,' forcing him to call his machine learning-based portfolio construction system by a different name — a restriction that persisted for many years inside Google.
Topics
Full transcript available for MurmurCast members
Sign Up to Access