Semiconductor processor maker Cerebras Systems revealed a massive $25.4 billion remaining performance obligation in its latest financial disclosures, driven largely by an expansive computing contract with OpenAI. The corporate update highlights both rapid expansion in artificial intelligence inference cloud services and significant revenue concentration risks for the chipmaker as it scales global infrastructure to meet unprecedented enterprise demand.
A Financial Milestone Fueled by Cloud Expansion
During the second quarter, Cerebras reported a 103 percent year-over-year surge in non-GAAP revenue, reaching $209.9 million as enterprise demand for specialized computing accelerated. The company's dedicated inference cloud business expanded nearly fourfold, prompting executive leadership to lift full-year revenue guidance to between $880 million and $890 million. These adjusted figures reflect strong momentum across high-performance enterprise workloads.
Standard accounting figures painted a similarly robust growth trajectory, with quarterly GAAP revenue climbing 74 percent year-over-year to $180.1 million. However, financial analysts tracking state regulatory filings point out that the company’s underlying valuation increasingly hinges on unfulfilled commitments. The $25.4 billion total backlog now stands at nearly 29 times the top-line revenue management projects for the entire 2026 fiscal year.
The Anatomy of the OpenAI Capacity Agreement
The extraordinary backlog stems primarily from a master relationship agreement negotiated with OpenAI in late 2025. Under the binding terms, the artificial intelligence developer committed to acquire 750 megawatts of specialized computing capacity dedicated to model inference, a deal valued at over $20 billion. Additionally, the agreement grants OpenAI an option to secure an extra 1.25 gigawatts of compute infrastructure through 2030.
Financial documents show that remaining performance obligations surged to $24.6 billion by the close of 2025 before edging up to $25.4 billion by mid-year. Because the vast majority of this capital was booked prior to the current fiscal year, organic backlog expansion slowed to roughly three percent over the past six months. Official filings confirm that OpenAI represents the core foundation of this future pipeline.
During the second quarter alone, Cerebras recognized $56.8 million in revenue under the OpenAI arrangement, accounting for roughly 32 percent of total recognized GAAP revenue. To support this massive build-out, OpenAI provided direct financial support in January through a $1 billion working capital advance facility. This cash injection underscored the strategic urgency of securing high-density computing assets in a constrained global hardware market.
Delivery Timeline and Structural Infrastructure Constraints
Translating billions in paper commitments into recognized revenue remains a multi-year engineering challenge constrained by physical data center construction schedules. Cerebras expects to recognize approximately 22 percent of the current backlog—or roughly $5.6 billion—over the 24-month period ending in mid-2028. Another 43 percent of the contracted revenue is scheduled for conversion between months 25 and 48, leaving the balance for later years.
Although this execution timeline spans several years, the near-term conversion rate has gradually accelerated compared to initial baseline estimates established last year. Management attributes the multi-stage rollout to the ongoing construction of specialized power and cooling facilities. Over 600 megawatts of data center capacity are currently operational or contracted for delivery by late 2027, paired with an aggressive tenfold manufacturing expansion.
To accommodate this aggressive scaling schedule, Cerebras is continuously broadening its geographical operations footprint across international energy hubs. The company recently announced plans for a 165-megawatt data center facility located in Finland, designed to capitalize on regional power stability and cold-climate efficiency. Industry analysts note that such infrastructure investments are vital for executing large-scale inference processing contracts without operational bottlenecks.
Persistent Customer Concentration and Market Risks
While the OpenAI partnership secures medium-term revenue visibility, it also highlights an ongoing pattern of extreme customer concentration. In 2025, official disclosures revealed that the Mohamed bin Zayed University of Artificial Intelligence generated 62 percent of corporate revenue, while Group 42 contributed an additional 24 percent. Second-quarter data shows three major clients still accounted for 76 percent of overall quarterly revenue.
Relying on a select cohort of enterprise clients creates structural vulnerabilities for investors, particularly if capital expenditure programs adjust rapidly. If key partners delay hardware deployments or modify their procurement schedules, revenue recognition timelines could experience significant disruption. Analysts emphasize that while heavy concentration provided rapid early growth, long-term stability requires diversifying the client ecosystem beyond a few mega-scale technology contracts.
Wall Street Outlook and Execution Challenges
Equity markets have displayed caution despite impressive top-line revenue growth and massive contract figures. Cerebras shares recently traded near $215, representing a decline of approximately 44 percent from their 52-week peak of $386.34. The valuation pullback reflects broader investor anxiety regarding capital intensity, data center execution risks, and the persistent reliance on high-profile anchor clients.
Moving forward, the primary metric for Cerebras will be its ability to systematically convert performance obligations into realized revenue while expanding manufacturing operations. Navigating supply chain constraints and completing data center installations on schedule will dictate whether the enterprise fulfills its record backlog. For now, the company’s financial trajectory remains inextricably linked to OpenAI’s long-term artificial intelligence deployment strategy.
