Cerebras Stock Is Up 11% Since Its IPO. Is It a Buy?

Cerebras (CBRS) surged at its $185 IPO on May 14 (opened at $350) but later faded to about $205, leaving late buyers ~10% underwater even though it remains ~11% above IPO. Core revenue grew 76% to $510M in 2025 and is guided +68%-70% to $855-$865M in 2026, supported by a $25B backlog, but gross margins are shrinking and the company is expected to stay unprofitable. The stock trades at 54x 2026 sales and ~6x projected 2028 revenue, with analysts expecting EBITDA to turn positive in 2027-28, keeping volatility elevated despite a large multi-year $20B OpenAI deal.

Analysis

The market is signaling that backlog alone is not a quality-of-earnings substitute. The key issue is conversion: if a large share of the pipeline is tied to customer-funded infrastructure or pass-through economics, headline growth can coexist with weak free cash flow and no durable margin expansion. That usually earns a lower multiple over time, especially for a newly public name with limited disclosure and a high single-name ownership base.

Competitive dynamics are more interesting than the article suggests. If Cerebras proves its inference architecture works, the first beneficiaries are likely not pure-play chip vendors but the surrounding infrastructure layer: AWS can monetize distribution, while NVIDIA may actually remain the structural winner because it captures the broader training/inference budget and has the software moat to absorb share shifts. The loser is the long-tail of AI hardware challengers whose equity stories depend on backlog being treated as recurring demand rather than a project-based order book.

Risk is asymmetric around the next two reporting cycles. In the next 1-3 months, the market will focus on gross margin trajectory, customer concentration, and whether revenue growth is outpacing capital intensity; any slip there can compress the multiple quickly. Over 6-18 months, the thesis only works if first-party infrastructure scales enough to reverse the current margin drag and if the OpenAI/AWS relationships translate into repeatable utilization rather than one-off deployments.

Key Decisions for Investors

• Do not chase CBRS on the current narrative; wait for one clean quarter of gross margin stabilization and cash conversion before underwriting the backlog. Falsifier: improving revenue without corresponding sequential gross margin expansion or operating cash flow.
• Pair trade: long NVDA / short CBRS on strength, 3-6 month horizon. Rationale: if alternative AI silicon matters, NVIDIA still owns the broader spend pool and software ecosystem; if it does not, CBRS rerates lower. Use a tight stop if CBRS demonstrates two consecutive quarters of margin inflection.
• Long AMZN as the lower-risk AI infrastructure beneficiary over 6-12 months. AWS can capture incremental AI inference demand and take rate without the manufacturing and capital-intensity burden sitting on CBRS.
• If borrow is available and liquidity permits, consider a small tactical CBRS short or put spread only after post-IPO volatility fades; avoid outright size until the next filing clarifies backlog quality versus pass-through revenue.