Groq Raises $350 Million at $3.5 Billion Valuation as Nvidia Deal Reshapes Its AI Infrastructure Strategy

Groq has raised $350 million in a financing round that values the AI infrastructure company at $3.5 billion, marking a sharp change from its previous valuation and business strategy. The round was led by Disruptive, with Nvidia set to participate, according to TechCrunch and Bloomberg AI. Groq is using the money to expand a neocloud business built around AI inference services and data-center capacity.

The valuation is about half the $6.9 billion level Groq reached in a $750 million round in September 2025. That earlier financing included BlackRock, Neuberger Berman, Deutsche Telekom Capital Partners, Samsung, Cisco, D1, Altimeter, 1789 Capital and Infinitum, according to Techfundingnews. The lower figure reflects a company reshaped by its agreement with Nvidia, rather than a simple continuation of its earlier chip-company model.

Nvidia’s relationship with Groq now sits at the center of the company’s transformation. Nvidia licensed Groq technology and hired founder and CEO Jonathan Ross along with other senior employees, according to TechCrunch AI and Bloomberg AI. The arrangement reduced Groq’s role as an independent AI-chip challenger while leaving it positioned to sell infrastructure and inference capacity.

Nvidia’s planned participation in the new round gives the chip giant a financial connection to Groq’s next phase. It also links Groq’s cloud expansion to Nvidia’s dominant accelerator ecosystem, as the startup builds data centers using Nvidia GPUs, TechCrunch reported. Nvidia’s involvement has attracted investor attention, while Barrons separately reported that analysts questioned the price attached to Groq technology and Nvidia’s stock slipped.

Groq is redirecting capital toward a neocloud model, supplying GPU-backed computing and AI infrastructure rather than relying primarily on sales of its own processors. The company’s focus is AI inference, the stage where trained models generate responses for users and software systems. Demand for fast, high-volume inference has become a major battleground as companies deploy conversational applications, agents and other AI products.

The shift places Groq in a crowded but expanding infrastructure market. Its earlier fundraising positioned it as a competitor to Nvidia in specialized AI chips, while the new financing emphasizes capacity, availability and service delivery. Groq’s stated goal is to expand its Nvidia-powered data-center footprint, giving customers access to inference infrastructure without building every system themselves.

Groq’s reset comes as other AI-chip startups continue raising capital and developing competing systems. SambaNova, an Intel-backed company, raised more than $350 million in a Series E round led by Vista Equity Partners through a partnership with Cambium Capital, Reuters reported through Siliconrepublic. Intel planned to contribute about $100 million, with commitments reaching as much as $150 million.

SambaNova’s financing was tied to a different strategy. The company introduced its SN50 processor for large-scale inference, announced deeper collaboration with Intel and secured more than $350 million to expand manufacturing and cloud capacity, according to HPCwire. The contrast shows how AI hardware companies are pursuing several routes at once: independent chips, strategic partnerships, cloud services and direct access to data-center buyers.

Groq’s $3.5 billion valuation puts a concrete price on the tradeoff between proprietary silicon and infrastructure scale. The company has raised substantial capital, but its new valuation is below the figure assigned before Nvidia’s licensing and talent deal. Investors are therefore assessing whether Groq’s cloud business can produce durable demand after its original chip strategy was compressed.

The financing also illustrates how Nvidia can influence competitors without simply eliminating them. By licensing technology, hiring key personnel and participating in Groq’s funding, Nvidia gains ties to an inference provider while Groq retains a route to market. Analysts’ questions about the technology’s price, highlighted by Barrons, add scrutiny to the commercial value of that arrangement.

Groq’s next test is execution: expanding data-center capacity, attracting inference customers and proving that a neocloud model can support its new valuation. The company’s access to Nvidia GPUs provides an immediate hardware foundation, but it also makes Groq’s growth partly dependent on the supply, economics and demand surrounding Nvidia’s platform.

The broader market will track whether customers favor specialist inference providers over direct purchases from cloud giants and chip vendors. SambaNova’s SN50 rollout and Intel partnership will provide another measure of demand for alternatives, while Nvidia’s participation keeps Groq tied to the incumbent it once challenged. Together, the deals point to an AI infrastructure market where chip design, cloud capacity and strategic alliances increasingly overlap.