By ByteBot
Humans&, a three-month-old AI startup founded by veterans from OpenAI, Anthropic, xAI, and Google, just raised a $480 million seed round at a $4.48 billion valuation. That makes it the second-largest seed round in startup history—106 times bigger than the typical AI seed round. The company has no product yet, 20 employees, and has existed for 90 days. Investors include Nvidia, Jeff Bezos, Google Ventures, and SV Angel’s Ron Conway. This is either supreme confidence in elite talent or spectacular late-stage FOMO.
The Numbers Don’t Make Sense—Unless They Do
Let’s put this in perspective. The median AI seed round in 2026 is $4.6 million. Traditional startups raise around $3.6 million. Humans& raised $480 million. That’s not just an outlier—it’s a different category entirely.
Only one seed round has been larger: Thinking Machines Lab’s $2 billion round, led by former OpenAI CTO Mira Murati. And even that story has a cautionary twist we’ll get to.
When a company with zero revenue, no launched product, and three months of existence gets a valuation 250 times higher than its peers, you’re witnessing one of two things: genius capital allocation or mass delusion. The founding team’s pedigree suggests the former. The market conditions suggest the latter. The truth is probably somewhere in between.
Elite Talent Doesn’t Guarantee Success
The Humans& founding team reads like an AI industry all-star roster. Georges Harik was Google’s seventh employee and led the Gmail launch, initiated Google Docs, and spearheaded the Android acquisition. Andi Peng worked on reinforcement learning for Claude 3.5 through 4.5 at Anthropic. Eric Zelikman and Yuchen He helped develop Grok at xAI. Noah Goodman teaches psychology and computer science at Stanford. The 20-person team includes researchers from OpenAI, Meta, and MIT.
This kind of talent concentration explains the investor appetite. There’s a massive brain drain happening across the Big Three AI labs. OpenAI employees are now eight times more likely to leave for Anthropic than the reverse, and Anthropic maintains an 80% retention rate compared to OpenAI’s 67%. Researchers are chasing autonomy, equity upside, and mission alignment.
But here’s the problem: elite pedigree doesn’t guarantee execution. Remember Thinking Machines Lab? Mira Murati’s $2 billion seed round with an even more famous founding team? Two of her co-founders—Barret Zoph and Luke Metz—left to rejoin OpenAI just months later. Talent is necessary but not sufficient.
Follow the Money—It’s All Going to One Place
The investor roster tells you everything about where AI capital is flowing. It’s concentrating in teams with resumes from OpenAI, Anthropic, and Google. If you’re a founder without that pedigree, good luck raising even $5 million in this market.
Nvidia led the round alongside SV Angel’s Ron Conway and co-founder Georges Harik. Nvidia has become the AI industry’s venture capital arm by necessity—it closed 67 AI deals in 2025 compared to 54 in all of 2024. The company committed up to $100 billion to OpenAI, $10 billion to Anthropic, and $2 billion to xAI. Critics call it “circular financing”—Nvidia invests in startups that use the capital to buy Nvidia GPUs.
Ron Conway’s participation is notable. His SV Angel fund backed OpenAI early and was described as “critical” to the company’s survival. His portfolio includes Hugging Face, Replit, Databricks, and Character.ai. When Conway writes a check this size, he’s not betting on a product roadmap—he’s betting on talent arbitrage.
The AI funding market is bifurcating hard. Seed funding activity dropped 29% year-over-year, but median valuations increased 19%. Fewer deals, bigger checks, all going to the same small circle of elite teams.
“Human-Centric AI” Needs Proof, Not Promises
Humans& positions itself around “human-centric AI”—tools that empower people rather than replace them, focused on communication and collaboration. The company plans to launch a product in “early 2026” but has released no specifics.
Here’s the issue: “human-centric” has become the 2026 equivalent of slapping “AI-powered” on every pitch deck in 2023. Microsoft Copilot claims to augment, not replace. Salesforce Einstein positions itself as an intelligent assistant that never replaces human judgment. Gartner predicts only one in ten agent interactions will be fully automated by 2026. Everyone pivoted to this messaging after the 2025 backlash against replacement narratives.
Without a concrete product, we can’t distinguish genuine differentiation from superior marketing. Humans& might be building the collaboration tools that reshape how developers work. Or it might be packaging familiar capabilities with better positioning. We’ll know when the product launches.
The $400 Billion Question
There are real bubble indicators here. 54% of fund managers say AI stocks are “bubbly”, and Lyft’s CEO bluntly stated “we are absolutely in a financial bubble.” The gap between AI infrastructure spending ($400 billion annually in 2026) and enterprise AI revenue (roughly $100 billion) is a 4-to-1 ratio that can’t persist indefinitely. Someone will get hurt when this corrects.
But Humans& isn’t raising on revenue projections. It’s raising on the thesis that an elite team from the top three AI labs can crack the “human-centric AI” problem before the market corrects. That’s a talent bet, not a product bet. Whether that’s genius or late-stage FOMO depends entirely on what ships in early 2026.
The infrastructure-revenue gap is real and concerning. But investors putting $480 million into a three-month-old company aren’t stupid. They’re making a calculated wager that when the AI market inevitably consolidates, the winners will be the teams with the deepest technical expertise and the strongest pedigrees. Time will tell if they’re right.
