Cerebras Systems vs. Nebius Group: Which Technology Stock Is a Better Buy in 2026?

For FY 2025, Cerebras reported revenue of about $510.0 million, up 75.7%, and net income of $237.8 million, but free cash flow was negative $392.8 million; Nebius reported revenue of about $529.8 million, up 350.9%, and net income of $101.7 million, alongside negative $3.7 billion free cash flow. The article favors Nebius for its faster growth, positive adjusted EBITDA, lower forward P/E (46.1x versus 217.4x), and infrastructure-financing partnerships, while noting its capital needs and competition from major cloud providers. Cerebras has proprietary wafer-scale chips and partnerships with OpenAI and AWS, but its shares are reported to be down more than 50% from their post-IPO peak, with concerns after reports that OpenAI selected Nvidia hardware for a newer product.

Analysis

The key distinction is not “chip versus cloud,” but where adoption and financing risk sit. Cerebras must earn durable developer adoption for a nonstandard stack; any design win can be valuable, but a customer choosing alternatives for a major workload would expose revenue concentration and weaken the case for premium sales multiples. Nebius shifts more risk to fleet utilization, power availability and the economics of financed capacity. Third-party funding can reduce upfront cash needs without eliminating fixed commitments or counterparty risk.

Both cases warrant a cash-quality discount until reported earnings reconcile with substantial negative free cash flow. For NBIS, track cash capex plus lease or other infrastructure commitments against incremental revenue and utilization—not adjusted EBITDA alone. For CBRS, verify customer concentration, repeat orders and cash collection; reports about a customer’s hardware choice are a catalyst to investigate, not proof of lost business. Confirm the periods, definitions and scope behind the article’s financial figures before using its P/E comparison.

Over 1–3 months, capacity delivery, utilization and customer disclosures should matter more than headline AI demand. Over 6–18 months, NBIS may gain if it converts scarce compute into attractive returns on committed capital; NVDA and, potentially, AMD benefit from continued accelerator demand, while hyperscalers Amazon and Microsoft retain scale and financing advantages. The contrarian risk is that scarcity-driven growth gets priced as durable economics before utilization and cash conversion are demonstrated.

Key Decisions for Investors

• Consider a modest, beta-adjusted NBIS-long/CBRS-short relative-value position only after confirming current filings and the next operating update. The thesis is stronger visibility into cloud demand versus greater adoption and customer-concentration risk at CBRS—not that NBIS is low risk. Reassess if NBIS capacity utilization or cash conversion disappoints, or CBRS demonstrates diversified repeat demand.
• Do not underwrite either name using headline P/E alone. Before adding exposure, reconcile net income to operating cash flow and free cash flow, and obtain NBIS’s infrastructure financing/lease commitments and customer concentration plus CBRS’s customer-level revenue and backlog. Treat missing disclosures as a reason to keep sizing small, not as evidence of safety.
• Monitor NVDA and AMD as potential beneficiaries of accelerator demand, but avoid treating NBIS growth as automatically incremental to either supplier: changes in sourcing, pricing and utilization can shift the economics. Falsify the bullish NBIS view if capacity comes online without utilization or if committed infrastructure costs rise faster than monetized compute; falsify the CBRS short thesis with evidence of repeat wins beyond a small set of customers.