Cerebras (CBRS) Q2 2026 earnings review
Cloud Hypergrowth Fuels Triple-Digit Gains, But Scaling Costs Compress Margins
Cerebras delivered a massive quarter with Core Revenue surging 103% YoY to $210M, driven entirely by a 287% explosion in its Cloud and Services segment. The company successfully executed a $6.4B IPO, eliminating any capital overhang and providing a massive war chest to fund its aggressive 600 MW data center build-out. With a staggering $25.4B Remaining Performance Obligation (RPO) backlog, the growth runway is heavily de-risked. However, management's celebration of YoY margin improvements masks a troubling sequential trend: aggressive scaling is severely compressing profitability, and the legacy Hardware segment is abruptly reversing course.
🐂 Bull Case
The transition from hardware vendor to cloud inference powerhouse is accelerating. Cloud and Services revenue nearly quadrupled YoY to $127.7M, surpassing hardware revenue and establishing a high-value, recurring revenue base.
An RPO of $25.4B combined with explicit management commentary planning to 'more than triple revenue in 2027' offers investors generational top-line visibility, largely anchored by frontier models like OpenAI's GPT-5.6 Sol.
🐻 Bear Case
Core Hardware revenue fell sharply on a sequential basis, dropping from $111.6M in 26Q1 to $82.1M in 26Q2. This signals a faster-than-expected cannibalization or lumpiness in physical system deployments.
Despite strong YoY metrics, sequential margins are breaking down. Core Operating Margin went from -2% in Q1 to -16% in Q2, with Q3 guidance pointing to further decay at -24%. The cost of 10x manufacturing scale is heavily front-loaded.
⚖️ Verdict: 🟢
Bullish. The sequential margin compression and hardware deceleration are real concerns, but they are symptoms of a company aggressively spending to capture a once-in-a-generation $25.4B backlog. The $8.6B liquidity cushion makes this aggressive investment phase highly sustainable.
Key Themes
Cloud & Services Now the Dominant Growth Engine
Cerebras is successfully executing a business model pivot. Core Cloud and Services revenue accelerated 287% YoY to $127.7M, making up 61% of total core revenue (up from roughly 41% last quarter). As data center capacity expands to 600 MW, this high-margin recurring revenue stream will act as the company's primary growth engine.
Contradiction: Margin Narrative vs Sequential Reality
Management boasted that they 'significantly improved core gross and operating margins compared to a year ago.' While technically true vs Q2 2025, it obscures a severe sequential deterioration. Core Gross Margin dropped from 46.5% in Q1 to 41.0% in Q2, and Core Operating Margin plummeted from -2.0% to -16.0%. Guidance implies further decay into Q3. The infrastructure ramp-up is temporarily breaking operating leverage.
Immunity to Macro Supply Chain Choke Points
A crucial macro advantage for Cerebras is its wafer-scale architecture, which inherently avoids the semiconductor industry's biggest bottlenecks. By bypassing the need for HBM memory, CoWoS advanced packaging, and 3nm fabrication—all of which are severely supply-constrained globally—Cerebras has de-risked its ability to scale manufacturing 10x in 2026.
Hardware Segment Reversing
While the company average growth was over 100%, the legacy Hardware segment is abruptly lagging. GAAP Hardware Revenue was $54.1M, down sharply from $70.3M a year ago. Even on a Core basis (adjusting for contra-revenue warrants), Hardware hit $82.1M—a massive sequential deceleration from $111.6M in Q1. Investors must monitor if this is a permanent structural shift to cloud, or if on-premise hardware deals are being delayed.
Disaggregated Inference & Tech Partnerships
Cerebras is overcoming inference throughput limits via strategic partnerships. By combining its chips with AMD and AWS (launching in Q4 2026 and Q1 2027, respectively) in a disaggregated inference architecture, Cerebras claims up to 5x throughput increases. Furthermore, securing the role of launch partner for OpenAI's GPT-5.6 Sol model validates its technological edge at the absolute frontier of AI.
IPO Distorts GAAP Profitability Metrics
GAAP metrics look abysmal on the surface—Operating Margin at (265%)—but this is entirely driven by IPO-related mechanics. The company took a massive $377M non-cash stock-based compensation (SBC) hit this quarter, up from $13.2M a year ago, alongside $22.7M in employer payroll taxes related to the IPO. The underlying Core metrics provide a much cleaner view of unit economics.
Other KPIs
An exceptionally high backlog relative to the current ~$885M annual revenue run rate. This essentially guarantees multi-year hypergrowth, supporting management's confidence in tripling revenue by 2027, provided they can scale infrastructure fast enough to recognize it.
Following the $6.4B IPO raise, combined with prior OpenAI loans and Series H funding, the balance sheet is pristine. The company also secured an $850M revolving credit facility. This removes any near-term funding risks associated with building out gigawatts of data center capacity.
Guidance
Decelerating sequentially. While this represents massive YoY growth, the QoQ growth rate will slow from ~10% in Q2 to ~2.5% in Q3. This likely reflects capacity gating as new data centers and manufacturing lines come online later in the year.
Accelerating into year-end. Subtracting Q1 ($191.3M), Q2 ($209.9M), and the Q3 midpoint ($215M) from the $885M full-year midpoint implies a massive Q4 Core Revenue of approximately $269M (a ~25% QoQ jump).
Deteriorating sequentially. This represents a further step down from Q2's (16%) and Q1's (2%). The company is absorbing heavy OPEX and R&D costs ahead of the late-2026 manufacturing and data center capacity explosion.
Accelerating/Improving into Q4. Because 1H margins were better, and Q3 is guided at roughly (24%), maintaining a full-year average of (18%) implies that Q4 margins must snap back significantly as the new $269M+ revenue scale finally delivers operating leverage.
Key Questions
Hardware Segment Outlook
With Core Hardware revenue down almost $30M sequentially, is this a permanent structural shift toward the Cloud segment, or were specific on-premise deployments delayed?
Operating Leverage Timeline
Core Operating Margins have degraded from -2% to an expected -24% in Q3. At what specific revenue utilization threshold do you expect these infrastructure investments to yield positive operating leverage?
RPO Concentration
Of the massive $25.4B Remaining Performance Obligation, how much is tied specifically to the OpenAI partnership, and what are the contractual safeguards around deployment timelines?