Cerebras Q2 2026 earnings beat and stock selloff

Cerebras' selloff starts with a reporting quirk most retail investors never saw coming: the company runs two revenue books. GAAP total revenue came in at $180.1 million, missing the roughly $194 million consensus estimate ^[1]. But Cerebras' own preferred metric, core revenue, hit a record $209.9 million, up 103% year-over-year and above the company's own prior guidance ^[1]. On profitability, the picture gets murkier depending on which outlet's numbers you use: SiliconANGLE reported an adjusted loss of roughly $0.05 per share that beat the $0.17 consensus estimate, alongside core revenue of $210 million that topped the $201 million forecast ^[5]. Yahoo Finance, by contrast, reported a loss of $2.98 per share that came in significantly wider than the $0.18 analyst estimate ^[2]. Those two loss-per-share figures are not measuring the same thing - they most likely reflect different bases (an adjusted, non-GAAP calculation versus a GAAP-approximate one) rather than a clean beat-then-miss on a single number - but the divergence itself shows how muddled the earnings picture was for investors trying to parse one clean signal. That bifurcation meant multiple, conflicting narratives were available on the same earnings print, and the market gravitated toward the harsher framing even as Cerebras itself called it "an outstanding quarter" with core revenue more than doubling and cloud revenue nearly quadrupling ^[3].

Beneath the headline numbers sits a more interesting mechanical story: core gross margin improved to roughly 40.6% year-over-year but fell sequentially from Q1's 46.5% ^[4]. Rather than reading this purely as weakness, Feldman framed it as a controlled trade-off - fast inference is priced at a premium and margins are "in a good spot and growing," with manufacturing efficiency and better component pricing expected to push margins back up going forward ^[5]. One YouTube analysis of the earnings call offered a more concrete mechanism for the sequential dip: it described Cerebras as renting back its own already-sold systems from hardware customers at a premium to keep pace with surging fast-inference cloud demand, with guidance pointing to Q3 as the low point for margins before a Q4 recovery. Read that way, the dip looks less like eroding unit economics and more like a deliberate, temporary capacity squeeze - a strategic choice that a headline-scanning market read as a red flag rather than the growth-over-margin bet it appears to be.

The most consequential shift in the quarter may be structural: cloud and services revenue rose 281-287% year-over-year and is now the majority of core revenue ^[1], a shift one YouTube analysis of the earnings call pegged more precisely at cloud/services jumping from roughly 43% of core revenue just one quarter earlier to about 61% in Q2. Hardware/product revenue, meanwhile, drew a starkly different reaction: The Information headlined the move as "Cerebras Shares Fall 16% On A Decline In Hardware Revenue" ^[6]. That framing captures the market's real anxiety - not that Cerebras is growing slower, but that its revenue mix is flipping toward a subscription-like cloud model faster than analysts had modeled. Notably, not every read of the underlying hardware trend was as negative: the same YouTube analysis argued core hardware revenue still grew a comparatively modest 17% year-over-year, held back mainly by "lumpy" delivery schedules tied to customers' data-center readiness rather than outright weakening demand - a more forgiving framing than the decline narrative that dominated coverage. Either way, the mix shift is strategically positive - cloud revenue is stickier and higher-margin over time - but it introduces lumpiness that unsettled analysts who had built their models around a steadier hardware-sale cadence ^[4].

Perhaps the clearest sign this was a sentiment-driven drop rather than a fundamentals-driven one: most sell-side coverage stayed bullish through the selloff. Morgan Stanley raised its price target to $279 and Citi maintained a Buy rating while trimming its target to $320 ^[7]; UBS went further, calling the drop a "compelling buying opportunity" ^[8]; even Mizuho, which lowered its target specifically on margin-pressure concerns, did not turn bearish ^[9]. TipRanks' consensus rating stood at Strong Buy on unanimous analyst support ^[10]. Meanwhile the decline was isolated to Cerebras - Intel climbed 6%, AMD rose 2%, and Super Micro Computer surged 11% the same day ^[3]- and Morgan Stanley highlighted Cerebras' own guidance implying core revenue more than tripling in fiscal 2027 ^[11], suggesting the stock move reflected repricing of near-term expectations rather than a reassessment of Cerebras' long-term trajectory.