Cerebras Systems Inc.

Executive Summary

SELL, Medium conviction. Rated SELL, Medium conviction for the next 6-12 months. The stock is already 68.2% below its 52-week high, but the decline has not made a loss-making company with 28.21x TTM P/S inexpensive relative to a 6.0x peer average. FY26 guidance and RPO support a strong operating trajectory, yet the valuation requires investors to underwrite a clean transition from rented systems and negative free cash flow to owned-system economics.

Investment Thesis

Avoid CBRS at $122.92 because the market is underwriting an exceptionally favorable sequence of events: data-center capacity must arrive on time, RPO must convert into recognized revenue, owned systems must reverse cloud-margin pressure, and concentration must remain manageable. Any one of those can work. Paying a premium for all of them before cash generation is established is the problem.

The operating evidence is stronger than the trailing financial statements suggest. Management delivered two consecutive reported EPS beats, 76.7% on August 12 and 75.6% on June 23. In the latest call, CEO Andrew Feldman said, "We delivered record core revenue and beat guidance on all metrics; core revenue, core gross margins and core operating margin." The FY2026 outlook was also raised, and CFO Robert Komin stated, "For the full year 2026, we're raising core revenue to the range of $880 million to $890 million. We're raising core gross margin to the range of 41% to 43%, and we're raising core operating margin to the range of negative 19% to negative 17%." This is a legitimate growth company with improving unit economics, not a broken business.

The issue is the bridge from promise to durable returns. Q2 EBITDA was -$0.43B, with a -236.5% margin, and management disclosed no adjustment. The margin model also gets worse before it gets better: Q3 gross margin is guided to 38%-40%, then management expects owned data centers to improve economics in Q4. That is a specific, testable claim, but it means the investment case currently relies on future asset deployment rather than demonstrated normalized profitability.

The stock's technical damage is notable, with shares 40.3% below the 200-day average and 36.99% down over one month. That creates scope for a squeeze because short interest is elevated, but it does not change the underwriting standard. A better entry requires evidence of Q4 margin recovery, clear conversion of the capacity pipeline into revenue, and a narrowing gap between growth and cash burn. Conviction remains Medium rather than High because those proof points can emerge quickly, but they are not yet in the reported numbers.

Key Metrics

Financial Strength

Liquidity is solid, with a 5.82 current ratio as of June 30, 2026 and modest 0.16 debt/equity, but financial strength is not the same as self-funded growth. Total debt of $1.5B and 5.32x FY2025 debt/EBITDA sit beside an operating model that consumed -$0.9B LTM FCF. Margin direction is operationally encouraging at the core level, but the consolidated LTM loss profile shows that capacity buildout, cloud economics, and corporate costs have not yet produced scalable cash earnings.

Competitive Position

Cerebras' moat is centered on wafer-scale compute, system-level integration, and demonstrated inference speed, rather than a conventional commodity-chip advantage. The company has embedded that technology in customer deployments and is pairing it with AMD Helios in a disaggregated inference design targeting 5x throughput. Durability is conditional: a >10x manufacturing expansion in 2026 can make the platform more available, but larger GPU vendors and hyperscalers can pressure performance, pricing, and software differentiation over the next 3-5 years.

Management & Guidance

Management has earned more credibility than a typical early-stage infrastructure company by beating EPS expectations in the two available 2026 observations and raising FY2026 revenue, gross-margin, and operating-margin guidance after Q2. The next test is unusually clear: Q3 revenue of $214M-$216M, gross margin of 38%-40%, and operating margin of -25% to -23%, followed by the promised Q4 gross-margin recovery as lower-cost owned systems arrive. Management's FY27 statement that core revenue will more than triple is ambitious, and credibility remains capped until capacity delivery and customer concentration are demonstrated in reported results.

Segment Analysis

• Core revenue (Not separately disclosed): Q2 2026 core revenue was $209.9M; management guides FY2026 core revenue to $880M-$890M.
• Cloud and data-center deployments (Not separately disclosed): Cloud deployment and customer usage ramps drove the Q2 core-revenue beat, but rented-back systems are pressuring near-term cloud margins.
• Geography (Not disclosed): No geographic revenue split was provided; AWS Bedrock availability expected in Q1 2027 is the disclosed enterprise-distribution expansion.

Capital Allocation

The balance sheet has a 5.82 current ratio and debt/equity of 0.16 as of June 30, 2026, which provides operating flexibility, but expansion remains capital intensive. The deep-dive feed reports $0.4B of capex and $0.4B of buybacks, with no dividends, while the financial dataset reports -$0.9B LTM FCF. That discrepancy makes the reported $0.5B deep-dive FCF figure difficult to reconcile; until cash reporting is clarified, capacity spending should take priority over repurchases.

Positioning & Flows

Ownership is meaningful but not dominant, with the top three institutions holding 35.9% and insiders holding 15.9%. Short interest is 23.20%, or 14.1M shares against a 61M float, with only 1.9 days-to-cover, so a positive earnings surprise can create a sharp technical rebound without changing intrinsic value. Options gamma is negative, but zero-gamma and call/put-wall levels were not supplied, so there is no defensible strike-level squeeze signal.

Bull Case

• Q2 core gross margin was 40.6%, about 940 bps YoY higher, and management lifted FY2026 core gross-margin guidance to 41%-43%.
• Q2 core operating margin was -16%, improving about 2,600 bps YoY and better than guidance; FY2026 guidance calls for -19% to -17%.
• Manufacturing capacity is targeted to expand more than 10x in 2026 through Flex and Sanmina, with 2027 facilities intended to add another 3x-4x.
• AWS Bedrock availability is expected in Q1 2027, adding a major enterprise route to market beyond direct deployments.

Bear Case

• LTM operating margin was -81.2% and LTM EBITDA was -$0.5B, leaving the earnings base far behind the valuation embedded in $28.8B market capitalization.
• The Q3 core gross-margin guide of 38%-40% is below Q2's 40.6%, because rented-back systems continue to dilute cloud economics before expected Q4 improvement.
• OpenAI is expected to remain a meaningful customer concentration in 2027, while the float has 23.20% short interest, raising both fundamental concentration risk and volatility risk.
• Data-center availability is management's primary growth bottleneck despite >600 MW of capacity, and the stated FY27 revenue ambition depends on facilities that are not fully operating today.

Valuation & Price Target

Engine price target $69.61 (-43.4% vs current), Low confidence.

• P/S Multiple: $14.62 (19% weight)
• Quality-adjusted: $122.92 (20% weight)

The conclusion aligns with the quantitative SELL baseline, though conviction is raised to Medium from the engine's Low confidence because the fundamental data now include a raised FY2026 guide, two large EPS beats, and $25.4B of RPO. The engine's $69.61 target, down 43.4% from $122.92, is reasonable as a base case because its sales-quality blend captures both the premium multiple and weak cash economics. The quality-adjusted method equals $122.92, showing why the stock can remain supported if the FY27 ramp lands, but the 3.1x peer-dispersion warning limits confidence in peer-derived valuation. Sell-side consensus is Strong Buy, with 8 buys and no holds or sells, but it is low-signal context and does not offset the valuation and cash-flow gap.

Risk Assessment

Risk score 50/100 (Moderate). The central risk is that the planned FY27 revenue acceleration fails to convert RPO into serviceable capacity, leaving a company valued at 28.21x TTM P/S with -$0.9B LTM FCF.

The Bottom Line

CBRS rates a SELL for 6-12 months, not because the AI demand story is weak, but because the stock still prices successful execution before profitability and cash conversion are established. The $69.61 base scenario recognizes real growth and a valid technology platform while assigning value to the risks in negative LTM cash flow and expensive sales valuation. The November 12 result could produce a sharp trading move given 23.20% short interest. The rating would change with sustained owned-system margin expansion, material reduction in cash burn, and evidence that the $25.4B RPO converts beyond concentrated customers.