Cerebras Systems, Inc.

Cerebras is not profitable on an underlying basis. The headline FY2025 GAAP net income (~$237.8M) is misleading, it was driven by a one-time, non-cash ~$363.3M gain from extinguishing a G42-related forward-contract liability. Strip that out and the non-GAAP net loss was ~$75.7M, wider than 2024's ~$21.8M loss.

The most important structural fact is concentration: in 2025 ~86% of revenue came from two UAE-linked entities (MBZUAI ~62%, G42 ~24%). The first quarterly print as a public company lands June 23, 2026. Unverified: exact gross / operating margin percentages were not confirmed from a primary source.

Growth is the headline draw: revenue +76% YoY, a $24.6B contracted backlog, and a $20B+ Master Relationship Agreement with OpenAI (750MW of capacity, expandable to 2GW, through 2028), with collaborations referenced involving AWS, Meta and IBM. The thesis is the shift from AI training to latency-sensitive inference, where Cerebras claims a large speed advantage.

Headwinds: extreme customer concentration, US export-control / geopolitical exposure tied to the UAE relationships, intense competition with NVIDIA, and a reported warning that revenue growth would slow in Q1 2026. No meaningful P/E or PEG given the underlying losses.

Classic Weinstein staging is not applicable, there is only ~4 weeks of public trading history, far too little to establish a 30-week base or trend. Instead, the post-IPO path: priced $185 → opened $350 → first-day close $311 (+68%) → intraday peak $386 → fell toward ~$200 → rebounded on analyst initiations → now consolidating in the low-$220s, last $219.18.

Any technical / stage-based criteria in the framework simply cannot be assessed responsibly yet for this name.

Long-term technical indicators (RSI, MACD, 50/200-day MAs, beta) do not yet exist for CBRS, there is insufficient trading history. What is measurable is extreme volatility: a +68% first day, a swing to $386 and back to ~$200, and live at-the-money implied volatility of ~118% (see §5).

There is no compression to analyze, quite the opposite. CBRS is one of the most volatile names in the market: live July at-the-money implied volatility is ~118% (call 117.8% / put 117.1%), and the stock has already round-tripped 80+ points since listing. No Bollinger structure or multi-month base exists yet.

Practically, ~118% implied volatility means the options market is pricing an exceptionally wide range of outcomes for the stock, a reflection of how much uncertainty surrounds a newly public, fast-growing, concentrated business.

Sentiment is broadly constructive following the post-quiet-period wave of new analyst coverage from 9+ major brokerages. The counterweight is a coming lockup overhang. Note: lockup share counts/dates are from a secondary source (Seeking Alpha) and are approximate, not confirmed to the prospectus.

Momentum into mid-June was strongly positive, driven by the simultaneous wave of bullish initiations (+11.4% then +18.1% on consecutive days around June 8). Today's −3.3% is a pullback within that. Cathie Wood's ARK reportedly bought ~$32M. With only weeks of history, momentum signals are event-driven rather than trend-based.

The macro tailwind is the AI capex super-cycle and the move toward inference, squarely Cerebras's pitch. The macro risks are unusually sharp for a single name: ~86% revenue concentration in UAE-linked customers intersects directly with US export-control policy, and NVIDIA's dominance is an ever-present competitive overhang. Unverified: current interest-rate level not gathered this pass.

Cerebras is a differentiated wafer-scale AI-compute company with rapid growth (+76%), a large contracted backlog ($24.6B), and a marquee OpenAI agreement, set against severe customer concentration, export-control exposure, underlying losses, a rich ~$52B valuation on $510M of revenue, and only a few weeks of trading history. Because the technical, stage, and compression criteria cannot yet be assessed on so little data, the analytical picture rests almost entirely on the fundamentals and the structural risks.

Supporting factors

Risk factors

What's working: the only commercially deployed wafer-scale processor, a first-mover in fast inference, +76% revenue growth, and a $24.6B backlog anchored by a $20B+ OpenAI agreement.

What could go wrong: ~86% of 2025 revenue from two UAE-linked customers; US-billed revenue reportedly shrank ~34% YoY; the headline GAAP profit is a one-time accounting gain masking a widening ~$75.7M underlying loss; export-control exposure; a rich valuation (~$52B on $510M revenue); a large lockup overhang; and ~118% option-implied volatility.

A note on the analysis: because the stage, long-term technical, and compression sections are un-assessable on ~4 weeks of data, any read on this company necessarily leans on its fundamentals and risks rather than its chart.