Cerebras stock falls 7% as OpenAI Ultrafast tier runs on Nvidia

By Splitfeed EditorialLast updated Oct 1, 2026, 07:10 UTC

On September 30, 2026, Cerebras stock dropped roughly 7% in a single trading session after reports emerged that OpenAI's newly launched GPT-6.1 Sol Ultrafast pricing tier runs on Nvidia chips rather than Cerebras hardware. The same day, Meta stock also declined, with analysts pointing to OpenAI's move as a competitive threat to Meta's Muse product. Semafor framed the situation as evidence of a broader compute bottleneck that OpenAI's new tiered pricing structure has brought into focus.

The market reaction reflects how concentrated Cerebras's perceived value had become around its relationship with OpenAI. Cerebras had been widely regarded as a key inference chip supplier to OpenAI, and that assumption was baked into its valuation. If OpenAI is routing its highest-demand, highest-speed workloads through Nvidia hardware instead, it raises real questions about the scope and durability of the Cerebras partnership. At the same time, Nvidia's positioning as the apparent beneficiary of OpenAI's Ultrafast compute demand reinforces a pattern that has played out repeatedly in this industry: during periods of peak demand, buyers tend to fall back on the most established supplier. What remains genuinely in dispute is whether this represents a one-off architectural decision or a signal that Cerebras is being sidelined in a strategically important product tier.

None of the three camps, Pro-AI, Anti-AI, or Middle Ground, had published reactions to this story by the time it was filed. Pro-AI voices would typically argue that Nvidia's role here confirms AI infrastructure is scaling as expected and that Cerebras still has room to compete in other tiers or with other customers. Anti-AI voices would likely use the story to highlight the fragility of AI hardware bets and the concentration of power flowing toward a small number of incumbent chip suppliers. Middle Ground observers would probably focus on the compute bottleneck angle, treating this as a sign that AI demand is outpacing diversified supply chains in ways that create market instability.

The clearest thing to watch is whether OpenAI or Cerebras offers any official clarification about the scope of their hardware arrangement, and whether Cerebras's next earnings report or investor communication addresses what share of its inference workload still flows through the OpenAI relationship. Any formal announcement about GPT-6.1 Sol's infrastructure would go a long way toward settling whether September 30 was a correction or a warning.