Cerebras Systems vs. Rigetti Computing: Is an AI or Quantum Computing Stock the Better Buy in 2026?

The article favors Cerebras Systems over Rigetti Computing, citing Cerebras's FY2025 revenue growth of 76% to approximately $510 million and net income of roughly $238 million, versus Rigetti's 34% revenue decline to about $7.1 million and $216 million net loss. Cerebras is projected by consensus to grow sales to $888 million in FY2026, $3 billion in 2027 and $7.5 billion in 2028, supported by AI inference-market growth expectations from $66 billion to $292 billion by 2029. Both companies carry substantial valuation and execution risk: Cerebras trades at 222x forward P/E and 71.2x sales, while early-stage Rigetti trades at 409x sales despite up to a $100 million federal R&D award.

Analysis

CBRS is not a clean AI-infrastructure analogue despite its revenue scale: its investment case hinges on converting a technically differentiated system into repeatable, diversified deployments. The key risk is customer and manufacturing concentration; wafer-scale economics can produce exceptional gross margins at utilization, but fixed supply commitments and deployment/service costs create severe downside operating leverage if a handful of large workloads shift toward NVDA-based clusters or hyperscaler ASICs. The relevant 1-3 month catalyst is evidence that backlog converts into multi-customer revenue and that cash burn narrows, not further upward revisions to distant 2027-28 estimates.

The market is likely underpricing the competitive response from AMZN, MSFT and GOOG. These buyers can partner with specialized inference providers while simultaneously steering internal workloads toward Trainium, Maia and TPUs; partnership announcements therefore do not necessarily establish durable vendor economics. This makes CBRS more vulnerable to a valuation reset than NVDA, whose ecosystem, software lock-in and broad customer base better support premium multiples. A deceleration in AI capex, lower inference pricing, or a material increase in CBRS working-capital needs would expose the mismatch between accounting profitability and negative free-cash-flow conversion.

RGTI should be treated as a duration-heavy R&D option rather than a revenue multiple story. Its liquidity extends runway but does not validate commercial demand; government funding with an equity component may also cap upside through dilution and create milestone-driven volatility. Over 6-18 months, the decisive variable is independently benchmarked error correction, fidelity and scaling progress relative to IBM and GOOG—not additional cloud-access partnerships, which have limited near-term monetization value.

Consensus framing of CBRS versus RGTI misses that both can decline together if speculative technology capital contracts. CBRS has a more credible operating business, but its implied growth path leaves little tolerance for execution misses; RGTI has greater binary upside only if a technical milestone changes the addressable market, an outcome not presently visible in financial results.

Key Decisions for Investors

• Do not initiate a directional CBRS long solely on AI-inference forecasts. Set an entry watch for the next earnings release: consider a small long only if management demonstrates customer diversification, sustained gross margin and a material improvement in free-cash-flow burn; exit on guidance reduction or renewed balance-sheet stress.
• Express AI infrastructure exposure through long NVDA versus short CBRS, sized beta-neutral, over the next 3-6 months. NVDA is better insulated from single-product and customer-concentration risk; cover the short if CBRS discloses durable multi-year contracted revenue, materially improves cash conversion, or the spread moves adversely by 20%.
• Avoid RGTI as a core long. For event-driven risk capital only, use a small defined-risk call position dated beyond the next major technical/government milestone rather than equity; maximum premium should reflect the high probability that commercial revenue remains immaterial over the next 12 months.
• Monitor AMZN, MSFT and GOOG AI-capex commentary and proprietary-silicon adoption. Any explicit shift of inference workloads toward internal accelerators is a near-term negative read-through for CBRS, while broad third-party accelerator capacity commitments would be the primary thesis validator.