Benchmark Commits $225M as Cerebras Valuation Jumps

Benchmark Capital has made an unusually large commitment to one of its earliest bets, raising $225 million through special-purpose funds to deepen its exposure to AI chipmaker Cerebras Systems.

The move follows a blockbuster financing for Cerebras this week. The company raised $1 billion in new capital at a $23 billion valuation, almost tripling its worth from the $8.1 billion valuation it achieved just six months ago. Tiger Global led the round, but a significant share of the capital came from Benchmark, according to a person familiar with the deal.

Benchmark’s involvement runs deep. The Silicon Valley firm first backed Cerebras a decade ago, leading its $27 million Series A round in 2016. Because Benchmark deliberately caps its core funds at under $450 million, it raised two separate vehicles, both called ‘Benchmark Infrastructure’, to support the latest investment, according to regulatory filings. The person familiar with the deal said the vehicles were created specifically to fund the Cerebras stake.

Benchmark declined to comment.

Cerebras has always stood apart by betting on physical scale rather than incremental design. Its flagship processor, the Wafer Scale Engine announced in 2024, measures roughly 8.5 inches on each side and integrates 4 trillion transistors into a single piece of silicon. The chip is manufactured from almost an entire 300-millimetre silicon wafer, the circular base material used across the semiconductor industry. Most chipmakers cut wafers into thumbnail-sized pieces. Cerebras uses nearly the whole disc.

That choice reshapes how AI workloads run. The Wafer Scale Engine contains 900,000 specialised cores operating in parallel, allowing calculations to stay on a single chip rather than moving data back and forth between multiple processors. That data movement is a major bottleneck in conventional GPU clusters. Cerebras says its architecture enables AI inference tasks to run more than 20 times faster than rival systems.

The funding lands as Cerebras builds momentum in the increasingly competitive AI infrastructure market. Based in Sunnyvale, California, the company last month signed a multi-year agreement worth more than $10 billion to supply 750 megawatts of computing capacity to OpenAI. The deal runs through 2028 and is designed to support faster responses for complex AI queries. OpenAI chief executive Sam Altman is also an investor in Cerebras.

Claims of superior performance place Cerebras in direct competition with Nvidia, whose chips dominate the current AI market. The question for customers mirrors one many businesses face when choosing between incumbents and challengers: does a radically different approach justify the operational risk of switching suppliers?

Cerebras’ route to the public markets has added another layer of complexity. Its relationship with G42, a UAE-based AI firm, weighed heavily on its prospects. G42 accounted for 87% of Cerebras’ revenue in the first half of 2024, but its historical ties to Chinese technology companies triggered a national security review by the Committee on Foreign Investment in the United States. That scrutiny delayed Cerebras’ IPO ambitions and led the company to withdraw an earlier filing in early 2025.

By late last year, G42 had been removed from Cerebras’ investor list, clearing a key obstacle. According to Reuters, the company is now preparing for a public debut in the second quarter of 2026.

Alongside the funding news, the broader technology ecosystem continues to mobilise around scale and execution. On June 23 in Boston, more than 1,100 founders are set to gather at TechCrunch Founder Summit 2026 for a day focused on growth and real-world scaling, underscoring how central infrastructure decisions have become to the next phase of the industry.

For Benchmark, the Cerebras bet reflects a familiar venture calculus. When conviction runs high and the market opportunity expands, firms stretch their own rules to stay close to potential category-defining outcomes. The scale of this commitment suggests Benchmark believes specialised AI hardware may still be early in its growth curve, with the biggest inflection point yet to come.

Author: George Nathan Dulnuan