Groq는 엔비디아의 참여로 35억 달러의 기업 가치에 3억 5천만 달러 규모의 시리즈 A 라운드 조달을 완료했다.
Groq, the AI-inference chip company that spent nearly a decade positioning itself as a scrappy alternative to Nvidia, has raised fresh capital at a valuation that quietly acknowledges how much the landscape has shifted.
On Monday, the company announced it had secured $350 million in a Series A round at a $3.5 billion valuation, roughly half the $6.9 billion it commanded last September.
The round was led by Dallas-based Disruptive, whose founder Alex Davis now serves as Groq’s executive chairman. In a development difficult to invent, Nvidia itself also participated.
Only months earlier, Nvidia had licensed Groq’s technology and recruited much of its engineering talent—an episode we covered as the company began rebuilding. Late last year, Nvidia struck a non-exclusive licensing agreement for Groq’s language-processing-unit technology. The deal was widely reported at around $20 billion and just as widely described as a “not-acqui-hire.”
There was no outright acquisition. Instead, Nvidia secured the intellectual property it sought and departed with founder and chief executive Jonathan Ross—a former Google engineer who helped develop the company’s tensor chips—alongside a significant portion of its senior leadership.
What remained required a new strategy and new leadership. Co-founder Doug Wightman stepped up as chief executive, a fresh bench of executives was recruited, and Groq repositioned itself. Rather than competing directly with Nvidia as a chip designer, the company pivoted toward operating data centres and selling AI inference by the token.
The stated ambition is now to build a “neocloud” that pushes capacity beyond 200 megawatts within a year. This inference-focused business—which already serves millions of developers and processes trillions of tokens weekly—was retained during the pivot rather than departing with the chip team.
A $650 million raise in June marked the opening phase of this reconstruction, with this $350 million round serving as the second tranche.
The optics of a company that once emptied its building now helping to refurbish it are peculiar, even by the standards of the AI-chip boom, where alliances are fluid and competitors frequently double as customers. This dynamic reflects a broader sector trend: Nvidia’s dominance has grown so complete that would-be challengers increasingly find partnership more viable than direct competition. Backing Groq costs Nvidia very little, yet secures a friendly, dependent supplier of inference capacity and a stake in the rebuilt firm’s future.
Still, the numbers warrant scrutiny. Halving a company’s paper valuation within a year represents a sharp correction in a market where valuations have largely moved upward, and where newer entrants continue securing billion-dollar price tags driven solely by inference demand. Rival inference-chip startups such as Fractile have raised at buoyant valuations, and one London-based challenger recently tripled its worth to $3.3 billion while openly positioning itself against Nvidia.
Groq’s discount serves as a reminder that gold-rush valuations price in founders and engineers as heavily as silicon, and that losing both carries a steep cost. Whether $350 million is sufficient to drive meaningful growth remains an open question. Groq still operates a genuine inference business, with data centres across multiple continents and developers who rely on it precisely because it is fast and cost-effective. Demand for inference is undeniable.
What remains uncertain is whether a rebuilt Groq—stripped of the founder who defined it and partially funded by the giant that hollowed it out—can become anything more than a comfortable, secondary supplier orbiting Nvidia. The valuation suggests investors have already accepted this narrower trajectory. At $3.5 billion, they are paying for a stable going concern, not a giant-killer.