Four chipmakers. One round. Zero products shipped.
Hark raised $700 million in a Series A at a $6 billion post-money valuation, confirmed across multiple outlets including a Reuters-syndicated report via TradingView. Parkway Venture Capital led the round. The co-investor list is where the story gets interesting: Nvidia, AMD Ventures, Intel Capital, and Qualcomm Ventures all participated, alongside ARK Invest, Brookfield, Greycroft, Align Ventures, Prime Movers Lab, Salesforce Ventures, and Tamarack Global.
That’s every major GPU and chip manufacturer simultaneously backing the same 70-person startup.
Hark was founded in late 2025 by Brett Adcock, who seeded it with $100 million of his personal capital. According to Let’s Data Science, citing company communications, Hark operates a data center equipped with Nvidia B200 GPUs and employs approximately 70 people. The company says it plans to release proprietary multimodal models in summer 2026, followed by native hardware devices. Adcock has stated the platform aims to deliver an AI that “actually knows you.” None of those capabilities have shipped publicly.
The pre-product valuation math: $6 billion post-money on a $700 million raise implies a $5.3 billion pre-money valuation. Against $100 million in founder seed capital, that’s a 53x step-up in roughly six months. Compare that to the broader 2026 AI unicorn cohort, where the median Series A for an AI hardware startup has come in at substantially lower pre-money multiples. The round is structured to reflect not current revenue but anticipated competitive position in a market that hasn’t yet been defined.
The catch is that the semiconductor investor composition tells a specific story. Nvidia, AMD, Intel, and Qualcomm don’t make coordinated bets on the same startup by accident. One interpretation: the chipmakers see Hark’s “universal digital interface” positioning as a potential distribution channel – a hardware layer that could sell through significant GPU and chip volume. Another: they’re hedging against a product category they don’t fully control yet and would rather own equity in the winner early than get locked out later. A third: Adcock’s execution track record (Figure AI, Archer Aviation, Vettery) is credible enough to attract strategic capital before the product is defined.
Probably some of each.
This is the third major AI hardware funding round in the current cycle following CoreWeave’s infrastructure raises and infrastructure-adjacent rounds documented across recent coverage. The pattern is consistent: capital is concentrating in the physical layer of AI, compute, chips, devices, not just the model layer.
Don’t read this as validation of Hark’s product. The company hasn’t shown the product publicly. The $6 billion figure reflects investor positioning, not demonstrated capability. What the round validates is the investment thesis: that whoever controls the personal AI hardware interface will have structural leverage over the model layer above it. The chipmakers believe that thesis enough to pay $5.3 billion pre-money to hold a seat at the table.
Watch summer 2026. Hark says its first multimodal models arrive then. That’s when the thesis meets the market.