The panelists debate the merits of Cerebras (CBRS) and Innodata (INOD), with most expressing concerns about their respective business models. While CBRS's wafer-scale architecture and OpenAI deal are seen as potential advantages, the company's cash burn and negative equity pose significant risks. INOD's high customer concentration and potential commoditization risks are also major concerns.
Risk: Cerebras' cash burn and potential dilution risk if milestones slip, and Innodata's 58% customer concentration and potential commoditization risks.
Opportunity: Cerebras' wafer-scale architecture and direct exposure to frontier-model scaling via its OpenAI supply deal.
This analysis is generated by the StockScreener pipeline — four leading LLMs (Claude, GPT, Gemini, Grok) receive identical prompts with built-in anti-hallucination guards. Read methodology →
Key Points
- Cerebras Systems produces massive wafer-scale chips that drastically accelerate the training and execution of complex artificial intelligence models.
- Innodata provides the critical data engineering and human expertise required by major technology firms to build reliable generative AI systems.
- Which high-growth artificial intelligence stock deserves a spot in your portfolio for the long term?
- …
Read more
Key Points
- Cerebras Systems produces massive wafer-scale chips that drastically accelerate the training and execution of complex artificial intelligence models.
- Innodata provides the critical data engineering and human expertise required by major technology firms to build reliable generative AI systems.
- Which high-growth artificial intelligence stock deserves a spot in your portfolio for the long term?
- 10 stocks we like better than Cerebras Systems ›
Artificial intelligence stocks continue to reshape the market in 2026 as investors hunt for long-term winners. Choosing between hardware powerhouse Cerebras Systems (NASDAQ:CBRS) and data specialist Innodata (NASDAQ:INOD) requires a look at two distinct strategies.
Cerebras designs massive, wafer-sized chips intended to power the next generation of artificial intelligence models. Meanwhile, Innodata provides the specialized human expertise and data engineering services that ensure those models are accurate and reliable. Both companies benefit from the same industry tailwinds, but their financial health and market valuations present starkly different paths.
The case for Cerebras Systems
Cerebras sells AI computing infrastructure, including its unique wafer-scale platform that uses a single massive chip to accelerate processing speeds. Its customers are spread across North America and Europe, focusing on high-performance deployments for research and specialized cloud services. In its latest official filing, the company reported having 708 employees as of late 2025 to support its global mission of making advanced computing more accessible.
In FY 2025, revenue reached $510.0 million, representing significant revenue growth of roughly 75.7% over the prior year. While Cerebras reported a GAAP net income of $237.8 million, this was entirely driven by a $363.3 million one-time, non-cash accounting adjustment, masking an actual operating loss of $75.7 million.
As of its December 2025 balance sheet, the debt-to-equity ratio was roughly -0.5x, which means total liabilities exceed shareholder equity. The current ratio, which measures a company's ability to pay short-term debts with short-term assets, was close to 2.1x. Free cash flow, or the cash left after paying for operating costs and capital investments, was negative $392.8 million for the fiscal year, though free cash flow equals cash flow from operations minus capital expenditures.
The case for Innodata
Innodata provides the data engineering and human expertise necessary for building large-scale generative AI systems for the world's largest semiconductor stocks and software giants. The company counts five of the "Magnificent Seven" among its clients, including Alphabet (NASDAQ:GOOGL) (NASDAQ:GOOG) and Amazon (NASDAQ:AMZN). However, one major customer accounted for approximately 58% of total revenue in FY 2025, and customer concentration like this adds a layer of risk to the business.
In FY 2025, revenue reached approximately $251.7 million, which is a revenue growth of nearly 47.6% compared to the previous year. Net income for the period was close to $32.2 million, resulting in a net margin of roughly 12.8%. While revenue increased, this net margin, which shows how much profit is kept from each dollar earned, saw a slight decrease from the 16.8% reported in FY 2024.
As of its December 2025 balance sheet, the debt-to-equity ratio was 0.0x, indicating the company has no traditional debt relative to its equity. The current ratio stands at approximately 2.7x, suggesting a healthy ability to cover upcoming bills with liquid assets. Free cash flow was nearly $35.6 million, though note that stock-based compensation represented roughly 23.8% of operating cash flow, which inflates reported cash generation since this is a non-cash expense.
Risk profile comparison
Cerebras faces intense competition in the AI hardware space from established giants such as Nvidia (NASDAQ:NVDA). The company must continue to innovate its wafer-scale technology to justify its high development costs and keep pace with rival hardware. Additionally, it faces risks associated with scaling its manufacturing processes and maintaining a consistent pipeline of enterprise customers to reach sustainable positive cash flow.
Innodata deals with extreme revenue concentration, as a single client provides more than half of its annual revenue and accounts receivable. It also navigates a rapidly changing technological landscape where it competes with massive firms like Accenture (NYSE:ACN) and Cognizant Technology Solutions (NASDAQ:CTSH). Geopolitical risks are also prominent, as the company operates in over 70 countries and faces legacy litigation in the Philippines that could impact its financial position.
Valuation comparison
| Metric | Cerebras Systems | Innodata | |---|---|---| | Forward P/E | N/A | 45.5x | | P/S ratio | 69.7x | 7.1x |
Valuation metrics sourced from Financial Modeling Prep (FMP) and may differ from other data providers.
Which stock would I buy in 2026?
I'd go with Innodata. While Cerebras is grabbing headlines with wafer-scale chips and OpenAI deals, Innodata has spent 12 consecutive quarters doing something harder to find in AI: growing consistently and becoming more profitable along the way. A broadening customer base keeps reducing its concentration risk, and the addition of a major frontier AI lab as a new customer adds a dimension to the story that most investors have not fully priced in yet.
Cerebras is doing technically impressive work, building inference chips that are faster than most alternatives. A multibillion-dollar supply deal with OpenAI signals serious institutional demand. Revenue nearly doubled year over year, and the company raised its full-year outlook after going public in May.
But Cerebras is still losing money, gross margins are shrinking as it scales, and the stock has dropped sharply since its IPO despite strong results. The pattern suggests investors are not yet convinced the business model holds up at scale.
For investors with patience and a long horizon, Innodata's proven execution and improving profitability make it the more comfortable place to put your money right now.
Should you buy stock in Cerebras Systems right now?
Before you buy stock in Cerebras Systems, consider this:
The Motley Fool Stock Advisor analyst team just identified what they believe are the 10 best stocks for investors to buy now… and Cerebras Systems wasn’t one of them. The 10 stocks that made the cut could produce monster returns in the coming years.
Consider when Netflix made this list on December 17, 2004... if you invested $1,000 at the time of our recommendation, you’d have $406,141! Or when Nvidia made this list on April 15, 2005... if you invested $1,000 at the time of our recommendation, you’d have $1,347,745!
Now, it’s worth noting Stock Advisor’s total average return is 940% — a market-crushing outperformance compared to 211% for the S&P 500. Don't miss the latest top 10 list, available with Stock Advisor, and join an investing community built by individual investors for individual investors.
**Stock Advisor returns as of September 18, 2026. *
Sara Appino has positions in Amazon and Nvidia. The Motley Fool has positions in and recommends Accenture Plc, Alphabet, Amazon, Innodata, and Nvidia. The Motley Fool recommends Cognizant Technology Solutions and recommends the following options: long January 2028 $260 calls on Accenture Plc and short January 2028 $280 calls on Accenture Plc. The Motley Fool has a disclosure policy.
AI Talk Show
Four leading AI models discuss this article
Opening Takes
“Cerebras could become the next big winner in AI hardware, but its profitability and execution risks mean the upside hinges on an unproven transition to wafer-scale adoption.”
The piece leans into a simple dichotomy: Cerebras is disruptive hardware, Innodata is a steady services play. But it glosses critical risks for Cerebras: a $363.3M one-time GAAP gain masked a $75.7M operating loss in 2025, negative free cash flow, and a balance sheet with negative equity that signals leverage risk should growth slow. The OpenAI deal is highlighted as proof of demand, yet it may be non-recurring or renegotiable. Also, Nvidia and other incumbents can erode any edge, and wafer-scale manufacturing has questionable yields at scale. Innodata’s 58% single-client risk remains a fragile pillar even if diversification improves.
Strongest counter: Cerebras may still win big if wafer-scale AI becomes mainstream, but that outcome relies on unproven scale economics and heavy capital burn; Innodata’s model already offers steadier cash flow, so this bullish bet hinges on an unproven tech trajectory.
“Innodata's extreme customer concentration and the threat of AI-driven synthetic data automation make its 45.5x forward P/E multiple fundamentally unsustainable.”
The article presents a false dichotomy between hardware innovation and data services, ignoring the precarious nature of Innodata’s (INOD) business model. A 58% revenue concentration with a single client is not a 'growth story'; it is a structural vulnerability that makes INOD a proxy for that client's shifting internal AI strategy. While Cerebras (CBRS) is burning cash, its wafer-scale architecture represents a genuine attempt at hardware differentiation against Nvidia's dominance. Paying a 45.5x forward P/E for a service provider like Innodata, which faces commoditization risks as AI models become more adept at self-labeling and synthetic data generation, is a dangerous bet on a 'middleman' business.
Innodata's 12-quarter growth streak suggests they have successfully embedded themselves as an indispensable utility in the AI supply chain, potentially justifying the premium through high switching costs.
“Innodata's customer concentration risk is materially understated; losing its 58% customer would halve earnings, yet the valuation prices in continued margin expansion as if that risk doesn't exist.”
This article presents a false choice. Cerebras' -$392.8M free cash flow and negative equity position are disqualifying for most investors—not a valuation debate. But the author's Innodata endorsement glosses over a critical vulnerability: 58% revenue concentration means losing one customer (likely the unnamed frontier AI lab) could crater profitability overnight. The 45.5x forward P/E assumes that customer sticks around and margins hold. Neither company is a comfortable hold; this is a 'pick your poison' scenario. Cerebras is pre-profitability hardware with real tech moat risk; Innodata is a services play with customer concentration that makes it a binary bet.
Innodata's improving margins and 12-quarter track record could signal genuine operational leverage as AI data work standardizes—and a 45.5x P/E isn't absurd if that frontier lab customer becomes 15-20% of revenue over 18 months rather than staying concentrated.
“Cerebras' OpenAI-validated hardware positions it for faster re-rating than INOD once utilization lifts gross margins above 40%.”
The article frames INOD as the steadier AI play due to profitability and lower valuation, yet underplays Cerebras' (CBRS) direct exposure to frontier-model scaling via its OpenAI supply deal and wafer-scale architecture. CBRS posted $510M FY2025 revenue with 75.7% growth and raised guidance post-IPO, metrics that historically precede margin inflection in hardware once utilization rises. INOD's 58% customer concentration and 23.8% SBC drag on cash flow create hidden fragility that service margins rarely offset long-term. Hardware leaders have repeatedly outpaced data-labeling peers during AI buildouts.
CBRS still burns $392.8M FCF annually with negative gross-margin trends at scale and faces Nvidia's entrenched ecosystem, so the OpenAI deal may prove non-exclusive and insufficient to reach sustained profitability before dilution hits.
The Debate
Responding to Grok
“Wafer-scale hardware is a binary, dilution-risk bet if milestones slip and OpenAI revenue isn't a durable moat.”
Responding to Grok: The OpenAI deal is a headline, not a moat. Even if CBRS achieves wafer-scale production, a 60-70% ramp in a few years assumes near-perfect yields, massive capex, and no competing fast followers. The risk is not just cash burn but timing: if profitability slips or the OpenAI revenue is non-exclusive or renegotiated, the upside compresses fast. My key claim: wafer-scale hardware remains a binary bet with dilution risk if milestones slip.
Responding to Grok
“Innodata faces an existential threat from synthetic data maturation, while Cerebras' growth is masked by unsustainable capital intensity and yield risks.”
Grok, you are conflating revenue growth with margin inflection. Hardware scale only benefits unit economics if yields stabilize, yet Cerebras remains in a high-capex, negative-equity hole. You discount the 23.8% stock-based compensation (SBC) at Innodata, but that is standard for growth-stage services. The real danger is the 'middleman' commoditization Gemini mentioned; if synthetic data generation matures, Innodata’s 58% concentration isn't just a client risk—it is an existential threat to their entire value proposition.
Responding to Gemini
“Innodata's commoditization risk is real but lagged; Cerebras' cash burn is immediate and more pressing for valuation.”
Gemini conflates two separate risks. SBC at 23.8% is indeed standard for growth services—that's not the issue. The real issue is whether Innodata's margin expansion (which Gemini acknowledges) survives if synthetic data commoditizes the core offering. But that's a 2-3 year tail risk, not imminent. Meanwhile, Cerebras' $392.8M FCF burn is happening now. Gemini's 'middleman' thesis assumes faster disruption than historical precedent suggests for entrenched data workflows.
Responding to Claude
“INOD faces faster commoditization risk than Claude allows, while CBRS hardware directly serves the same customers.”
Claude's 2-3 year timeline for INOD commoditization ignores how quickly OpenAI-scale labs could deploy synthetic data tools to cut external labeling spend. That would turn the 58% concentration into an immediate hit at 45.5x forward P/E, not a distant tail risk. CBRS cash burn is real, but its wafer-scale exposure to the same labs creates a nearer-term offset if utilization ramps, a linkage the panel has not stress-tested.
Panel Verdict
NEUTRAL No ConsensusThe panelists debate the merits of Cerebras (CBRS) and Innodata (INOD), with most expressing concerns about their respective business models. While CBRS's wafer-scale architecture and OpenAI deal are seen as potential advantages, the company's cash burn and negative equity pose significant risks. INOD's high customer concentration and potential commoditization risks are also major concerns.
Cerebras' wafer-scale architecture and direct exposure to frontier-model scaling via its OpenAI supply deal.
Cerebras' cash burn and potential dilution risk if milestones slip, and Innodata's 58% customer concentration and potential commoditization risks.
This is not financial advice. Always do your own research.