Cerebras Systems (NASDAQ: CBRS) and NVIDIA (NASDAQ: NVDA | NVDA Price Prediction) delivered post-earnings reports showing two distinct AI compute visions. NVIDIA reported a $81.61 billion quarter powered by full-stack AI factories. Cerebras, fresh off its IPO, bets on wafer-scale inference speed. The results make a direct comparison unusually relevant.
Wafer Scale Meets AI Factory Scale #
NVIDIA’s Q1 FY27 earnings were dominated by Data Center revenue of $75.246 billion, up 92% year over year, with networking growing 199%. Jensen Huang stated “Demand has gone parabolic.” Blackwell Ultra swept every MLPerf inference benchmark, and Vera Rubin production begins in the second half of fiscal 2027.
Cerebras took a different route. Core revenue reached $209.87 million, up 103%, though GAAP revenue of $180.11 million came in 6.95% shy of consensus. CEO Andrew Feldman said “our cloud business nearly quadrupled year-over-year”, driven by fast inference demand from OpenAI, AWS, AMD, and CrowdStrike. Backlog sits at $25.4 billion in remaining performance obligations.
Platform Empire Versus Speed Specialist #
The strategic split is clear. NVIDIA sells a vertically integrated stack: CUDA, NVLink, Spectrum-X, DGX, and Vera Rubin, aimed at every hyperscaler and sovereign buildout. Huang emphasized that “AI native clouds don’t build chips, don’t design their own chips”, positioning NVIDIA as the ready-made AI factory. Cerebras leverages single-chip scale that sidesteps HBM memory, CoWoS packaging, and 3nm constraints, wrapped in a cloud inference service.
Lens | NVIDIA | Cerebras | | Core Bet | Full-stack AI factories | Fastest single-user inference | | Gross Margin | 75.0% non-GAAP | 40.6% core | | Anchor Customer | Every hyperscaler | OpenAI 750MW deal | | Key Vulnerability | China revenue at zero | Customer concentration |
NVIDIA counters Cerebras with TensorRT-LLM, speculative decoding, multi-user concurrency scaling, and ubiquitous cloud availability. Cerebras wins on raw tokens per second for a single user. Different buyers, overlapping budgets.
Vera Rubin Ramp Versus Cerebras Scaling #
The next tests are concrete. NVIDIA guided Q2 FY27 revenue to $91.0 billion, plus or minus 2%, excluding China Data Center compute. Vera Rubin promises up to 35x higher inference throughput versus Blackwell. Watch whether hyperscalers accept that leap without hesitation.
Cerebras raised full-year 2026 core revenue guidance to $880 to $890 million and plans to triple revenue in 2027. Monitor whether manufacturing capacity, scaling more than 10x in 2026, actually delivers.
Why NVIDIA Remains the Anchor, With Room for Cerebras #
NVIDIA remains the sturdier position. A 32x trailing P/E on 17.28% one-year gains, paired with $48.554 billion in quarterly free cash flow, reflects a business printing money while buildout is early. Cerebras appeals as a smaller, higher-variance position for investors believing inference speed commands a premium. GAAP losses tied to $377.0 million in stock-based compensation and customer concentration warrant caution. If Vera Rubin ships on time and Cerebras hits 2027 guidance, both can work. Until then, NVIDIA is the anchor and Cerebras is the swing.
Contact [email protected] for any questions or corrections.