Roger Luo

Every day there are panicked articles about the carnage on white-collar jobs that AI is wreaking.

How AI has destroyed the value of young people's college educations.

Those engineering degrees? Worthless.

Is any investment in education safe in this new AI era?

Well, maybe one. Maybe that of Roger Luo.

See, Roger got a PhD in AI back in 2011. About fifteen years before that became the single most valuable skill set on the planet. It's certainly paying off now.

Roger recently closed on a new $16 million fund for a new firm called Embedding VC. It took him two years. But that's not a long time when you consider he often spends two years working with a potential new AI company before he invests. Not because he's indecisive, but because these companies can be so bleeding edge that the founder might spend months or years with Roger, trying to decide if it's even viable first.

If there were such a thing as a "full stack" AI expert, it would be Roger. He worked in academia, his inner introvert sated by nary a sales or pitch deck. He worked in AI at public companies like Yahoo and Snap, where he was a founding member of Snapchat Research, developing machine learning and early AI efforts. He co-founded a YC company, Mosaix.ai, which was acquired by AWS in 2020, and then co-founded Eve, an AI legal tech company that's now a unicorn. He's been a part-time venture partner, Sequoia Capital scout, and an angel investor for nine years, investing in 35 deals. Among them there are three IPOs and nine other exits.

But all of those bragging rights, resume bullet points, and impressive proper nouns aside, the reason limited partners wanted to invest in Embedding is because AI investors seek him out well before they've started an AI company. When they are thinking about whether they are going to, because of this expertise.

He's an AI founder's AI investor.

(You want to see what a true AI fund's homepage looks like? Go to Embedding's site. You have to type "exit" to get to the non-terminal, regular web version. It's not for the normies.)

Clearly, Roger is someone who bet his career on AI. But it wasn't a calculated mercenary move. When I asked him what's surprised him about AI going back to a decade and a half ago when he got his PhD, it was an easy answer: How fast it's moved and how big it's gotten.

"What I once predicted would have taken decades to achieve has happened in a so much shorter time frame," he says.

That kind of humility and curiosity keeps him grounded and engaged in the future of AI, not just what can make money. He believes we've barely scratched the surface of the opportunities when it comes to companies that revolutionize verticals like education and legal tech, that democratize healthcare and predict future trends.

He does roughly ten deals a year, writing pre-seed checks up to $300K and seed checks up to $1M.

Just like founders have come to him through networks and word of mouth, so too did LPs. Like most emerging VCs, he started raising money through friends and friends of friends. Word spread as he started investing in companies, he says. His LPs share his engineer-focused bent. This shop is deep in the AI trenches.

Roger's fund echoes all the trends we see of funds that outperform in challenging markets:

Highly specialized expertise. Roger's PhD in AI and decades of hands-on experience give him an edge that generalist funds simply can't replicate.

Individual LPs, some of which may be new to the asset class, that know how to value that expertise. His investors understand what they're backing and why it matters.

Funds that are so early they are effectively operating as institutionalized friends and family rounds. Embedding works with founders before they've even decided to start a company.

Venture capital has become completely bifurcated, and it's in no small part because of AI. Mega funds have become even bigger, pouring more and more billions into the data center costs of AI and hoping to reap massive returns when a few winners go public. These $5 billion, $10 billion, even $20 billion funds can't rely on AI alone to get returns on those massive funds.

They likely need to invest in other deep tech spaces like energy, robotics, and space. But a $5 billion fund can't roll up its sleeves and operate as the discovery layer capital for companies like those. Neither can generalist pre-seed funds.

It takes funds like Embedding and investors like Roger Luo