The one number
$25.4 billion of remaining performance obligations.
That is the future contracted revenue Cerebras reported with its June-quarter results. It is a demand signal, not cash received, profit earned or next year's revenue. The investment question is how quickly the company can supply that compute, and what shareholders retain after the cost of delivery. August results and outlook.
Cerebras is today's Stock of the Day for the first time. Its technology and commercial progress deserve attention. Its stock also needs a separate test: a lower share price can reduce a valuation multiple without making the expectations easy to meet.
The inference opportunity
Cerebras builds wafer-scale AI computing systems and sells both hardware and cloud services. Inference is the computation that generates a model's responses. Faster responses can make repeated, multi-step AI tasks more practical. The commercial test is whether customers pay enough for that advantage to cover the infrastructure and financing involved.
The August CS-4 launch illustrates the product ambition. Cerebras advertised up to 30 times the speed of GPU-based solutions in specified comparisons. Those are company claims tied to workloads and configurations; they are not evidence of the same advantage in every application or a forecast of market share. CS-4 announcement.
September's Finland announcement shows the infrastructure side: a planned 165 MW campus, with an initial 50 MW construction phase and seven-year service orders. Full planned capacity and capacity already serving customers are different measures. Neither headline alone tells us utilization or profit per megawatt. Mikkeli project.
More cloud revenue; two different margin stories
The earlier quarter matters. Q1 GAAP revenue was $193.4 million, including $82.8 million of cloud and other services. Q2 cloud revenue increased to $126.0 million, but total GAAP revenue fell to $180.1 million, down 6.9% sequentially. Fast cloud growth and a lower total can both be true. Q1 income statement; Q2 results.
The graphic compares Q1 and Q2 across both years; it deliberately leaves out Q3 and Q4 rather than inventing a continuous quarterly series. Gross margin should be read beside the revenue definition. GAAP and company-defined core results answer different questions.
Read the reconciliation before the growth headline
Q2's $209.9 million core revenue starts with $180.1 million GAAP revenue, removes $14.5 million pass-through revenue and adds back $44.3 million customer-warrant amortization. The revenue bridge is shown explicitly so readers can see what changed.
Core results help examine the business without specified accounting effects. They do not make equity compensation or customer warrants economically irrelevant. A non-cash charge can still matter to the portion of future earnings belonging to an existing share.
The June-quarter operating loss was $477.2 million under GAAP versus $33.6 million on the core basis. Both were losses. This is a comparison of accounting measures, not a cash-burn waterfall. Reconciliation tables.
What the price is asking the company to deliver
At $181.55, multiplying by the 237.564 million Class A, B and N common shares reported as of August 5 gives an approximately $43.1 billion equity-value proxy. All three classes participate economically. This is a dated share-count calculation, excluding unexercised awards and later issuance, rather than a verified current market capitalization. 10-Q share-count cover; share-class rights.
Against the $885 million midpoint of management's August FY2026 core-revenue outlook, that proxy is 48.7 times forecast core sales. This is neither P/E nor enterprise value divided by GAAP sales. The history starts after the August outlook was released and holds its share count and revenue denominator fixed; it illustrates price movement against those assumptions, not changing historical analyst consensus.
At the same price and share count, hypothetical annual core revenue of $2.655 billion would reduce the ratio to 16.2 times. That does not make the stock automatically cheap: revenue must be delivered with sufficient margins, and financing and dilution can change the numerator. A smaller sales multiple cannot tell us how much profit arrives per share.
Rather than forcing a positive P/E out of a quarterly loss, this review makes the revenue assumptions visible. Valuing future earnings would also require an explicit profitability path, share count and time horizon.
Funding helps; it does not finish the execution
June 30 cash and equivalents were $6.742 billion, with $1.179 billion investments and $684.7 million restricted cash separately reported. The company also had $918.2 million working-capital borrowing. First-half operating cash use was $47.5 million and property-and-equipment purchases were $548.9 million. Cash availability, accounting loss and infrastructure investment are separate measures. 10-Q balance sheet and cash flows.
The remaining question is how effectively that capital becomes revenue-producing capacity. Large reserves reduce an immediate funding constraint; they do not remove construction, utilization, customer concentration or competitive risk. Progress should be assessed through deployments, customer consumption, margins and cash generation together.
A strong Monday bounce is an observation, not a completed recovery
Cerebras closed October 5 at $181.55, up 9.1% that session. It still sat 9.5% below the 50-day average of $200.70. Regaining that stationary reference would require about 10.5% upside; the actual average continues moving. Daily bars used for calculation.
The supplied October 6, 06:43 ET premarket mark is $183.03, about 0.8% higher, and is not a live quote. The chart above redraws the independently checked closing bars since the IPO and keeps volume in its own panel. The supplied premarket chart below uses a shorter July-to-October window.
Cerebras began trading on May 14, leaving 99 completed sessions through October 5. Standard 100-day and 200-day averages are not available on that history. The $185 IPO offer price is a different reference from an exchange close or anyone's actual purchase cost. IPO announcement.
Higher lows, sustained improvement through recovered trading ranges and stronger relative performance would provide more price evidence. A rising business metric and a rising stock on one morning do not establish the same thing.
The next checkpoint
Our interpretation: Cerebras has a credible business opportunity in fast inference. Its contracts and technology need to become durable cash flow, while the share price needs to sustain the repair that Monday's rebound began to suggest.
Watch capacity entering service, revenue recognition and customer concentration. Then read GAAP and core results together and revisit the share count. The attractive story becomes a stronger stock case when those operating developments and price behavior begin to agree.
The question: can fast inference become profitable growth fast enough to justify the expectations still embedded in the shares?
Educational commentary, not a buy or sell recommendation. Regular-session basis: October 5, 2026. Premarket observation: October 6, 06:43 ET. Reported financials: quarters ended March 31 and June 30; guidance remains the separately dated August outlook.