By James Aspinwall
Dave Blundin has been building neural networks since the 1980s, long before “AI” meant anything to most people. While the rest of the venture world runs on gut instinct and warm introductions, Blundin built a machine that uses five billion rows of consumer data per day to decide what companies to create. He’s co-founded 23 companies, at least five hit nine-figure valuations, and he teaches AI entrepreneurship at MIT. He also calls himself the “Godfather of Quantization” – and given that quantization is now foundational to every major LLM deployment pipeline, the title isn’t entirely tongue-in-cheek.
MIT AI Lab to MicroStrategy
Blundin studied Computer Science at MIT, where he researched neural network technology at the CSAIL AI Lab. He was among the first ten employees at MicroStrategy, Michael Saylor’s analytics company (now rebranded as Strategy, the Bitcoin treasury company).
In 1992, he pioneered the quantization of neural networks at Scan-Optics, deploying them on custom inference-time hardware for optical character recognition, demand forecasting, and medical imaging. This was 30 years before quantization became the standard technique for deploying large language models efficiently. MIT Technology Review named him one of their “Innovators Under 35,” and Duncan McCallum of One Liberty Ventures called him “the most visionary person I have ever met” in decision-support software.
His personal philosophy, stated in college: “I’d rather be destitute trying to achieve greatness than mediocre trying to achieve mediocrity.”
The Billion-Dollar Exit
In 1994, Blundin founded Cirrus Recognition Systems, applying the pattern recognition software he’d developed at MIT. Then in 1997, he founded DataSage, an early data mining company that used neural networks to analyze terabytes of clickstream data for Amazon, Walmart, and major financial services firms.
In 2000, Vignette Corporation acquired DataSage for over $1 billion on a cash-adjusted basis. Blundin stayed on as Chief Technologist. At 32, he’d already built and sold a unicorn before most people in the current AI wave were in middle school.
Cogo Labs: The Data Factory
After a stint on MicroStrategy’s board (2003-2004), Blundin co-founded Cogo Labs in 2005, located next to MIT in Cambridge. This is where his approach diverges from every other venture firm.
Most incubators wait for founders to show up with ideas. Cogo Labs generates the ideas from data. The operation collects over five billion rows of consumer data per day and maintains multi-petabyte datasets. Proprietary algorithms identify business opportunities, internal teams run experiments on Cogo’s platforms, and profitable ventures get spun out as independent companies.
Through this process, Blundin became chairman of 12 companies. Notable spin-outs include EverQuote (the largest online auto insurance marketplace, IPO’d on NASDAQ in 2018), Autotegrity (sold to ADP in 2012), CourseAdvisor, and Datalign Advisory.
The key insight: “Link’s investment strategy relies not on subjective preferences but rather on the objective identification of markets and companies as revealed by the data.” In a world where most VCs make decisions based on pattern matching and personal networks, Blundin uses actual patterns in actual data.
The Link Ventures Ecosystem
Blundin founded Link Ventures in 2006, and it’s grown into something more than a fund – it’s a full-stack venture ecosystem:
Link Ventures – Cambridge-based AI venture firm backing elite technical founders from MIT, Harvard, and beyond. Link Ventures 3 LP closed at approximately $150 million in 2023, 50% larger than the previous fund. Total AUM across all funds exceeds $500 million.
Link-XPV (Exponential Ventures) – A joint fund with Peter Diamandis focused on AI startups. About 75% of deals come from the MIT/Harvard ecosystem. KKCG, Karel Komarek’s investment group, has invested in this fund.
Vestigo Ventures – A fintech-specific fund where Blundin is General Partner. It uses exclusive access to Cogo Labs’ data and proprietary ML models to predict which fintech companies will gain traction.
Link Studio – A 40,000-square-foot incubator in Cambridge that provides capital, compute resources, operational support, and even living space for founding teams.
Vestmark – Co-founded around 2001 with Chuck Johnson. Blundin served as Chairman and founding CEO. The wealth management technology platform now manages over $1 trillion in US assets with 1.6 million accounts.
Notable portfolio investments include Liquid AI (MIT spinout, $2B unicorn as of December 2024), Healthcare.com, Canela Media, JobCase, Trust & Will, and Caribou.
The Moonshots Connection
Blundin sits on the XPRIZE Foundation board and is investment partners with Diamandis through Link-XPV. He’s not an occasional podcast guest – he’s a regular co-host on Moonshots with Peter Diamandis, appearing alongside Salim Ismail and Alexander Wissner-Gross in episodes covering AI, deflation, geopolitics, and market dynamics.
On the podcast, Blundin brings the technical depth that comes from decades of building neural networks, not just investing in them. He’ll reference his 1992 quantization work in the same conversation as Gemini 2.5 Pro’s latest capabilities. That historical perspective is rare in a space where most commentators discovered AI in 2022.
What He Actually Thinks
Blundin’s views are sharp and specific:
On superintelligence: He predicts an intelligence explosion in 2026 or 2027. Not a gradual improvement – a step function producing 100x to 10,000x improvement in software capability. He accepts Elon Musk’s ASI timeline: “I would not doubt his timeline.”
On the AI bubble question: “Absolutely not a bubble. It’s the biggest shift in human history.” He acknowledges many AI companies will fail – “all kinds of charlatans running around raising capital” – but the underlying technology shift is real.
On 40x deflation: He warns about sustained year-over-year hyper-deflation in the cost of intelligence dragging down the price of everything else. He’s genuinely concerned about inequality: “The wealth isn’t going to go to the people doing the work or the people who get unemployed… it’s going to go to making the rich richer and the poor poorer.”
On education: The current curriculum is “not even vaguely useful” given the AI revolution. College students with startup ambitions should consider dropping out now.
On longevity: He believes it makes sense to “double down on your health” because we’re likely to reach “mortality escape velocity.”
On investing: Talent and computational resources now limit growth more than capital. He views Cursor (the AI code editor, $500M ARR in three years) as the model for future AI startups.
The MIT Classroom
Blundin teaches “AI for Impact: Venture Studio” at MIT, a course that’s “vastly oversubscribed” with about 100 students. Co-taught with Media Lab professor Ramesh Raskar, it uses a “spot-probe” framework: students identify opportunities using data signals, then rapidly validate solutions. It’s the Cogo Labs methodology, distilled into a curriculum.
The Bottom Line
What separates Blundin from most AI investors is the depth of his technical history. He’s not a finance person who pivoted to AI when it became fashionable. He was quantizing neural networks when Bill Clinton was in his first term. He built a $1 billion exit on data mining before the term “big data” existed. He constructed a venture ecosystem that generates deal flow algorithmically rather than socially.
The full-stack pipeline – Cogo discovers opportunities through data, internal teams validate them, Link invests and scales them, Vestigo captures the fintech vertical, Link Studio houses the founders, and MIT provides the talent – is a genuine moat. It’s not the most famous venture firm, but it might be the most systematically constructed one.