Nvidia CEO Jensen Huang rejects talk of AI bubble: 'We see something very different'

Nvidia reported third-quarter revenue and profit that topped estimates, issued stronger-than-expected guidance and said its order backlog — now including recent deals with Anthropic and an expanded Saudi agreement — supports CEO Jensen Huang’s $500 billion sales outlook for calendar 2025–26. Huang dismissed talk of an “AI bubble,” arguing GPUs will supplant CPU-based infrastructure for data processing, enable entirely new applications and power emerging “agentic” AI that will demand substantially more compute, and positioned Nvidia as uniquely able to serve those three use cases. The results and commentary aim to allay investor concerns about hyperscaler concentration, elevated capex and external financing (which management says is the customers’ responsibility), signaling continued robust demand for Nvidia chips even as other AI-linked stocks have pulled back.

Analysis

Nvidia reported third-quarter revenue and profit that exceeded estimates and issued better-than-expected guidance, with CEO Jensen Huang rejecting assertions of an AI bubble and noting the company has grown to a $4.5 trillion market cap on soaring GPU demand. Huang articulated a three-pronged rationale for durable demand: GPUs displacing CPU-based infrastructure for data processing and recommendations, enabling entirely new applications, and powering "agentic" AIs that will require substantially more compute. Management highlighted a $500 billion sales forecast for calendar 2025–26 and said the order backlog — now including a newly announced Anthropic deal and an expanded Saudi agreement — does not yet reflect recent orders.

Huang stressed Nvidia’s broad visibility into demand because every major cloud provider (Amazon, Microsoft, Google/Alphabet, Oracle) and leading model developers (OpenAI, Anthropic, xAI, Meta) are customers, and CFO Colette Kress said the company is on track to hit its sales forecast. Nvidia acknowledged investor concern about hyperscaler concentration and external financing used by customers to build infrastructure, but management argued those chips increase hyperscaler revenue by powering recommendation systems and other monetizable services. Prior to the release NVDA shares were down ~8% for the month while other AI-linked names fell sharply (CoreWeave -44% in November, Oracle -14%, Palantir -17%), indicating investor rotation and risk repricing.

Key implications are clearer demand visibility and stronger-than-expected near-term fundamentals for Nvidia, balanced by ongoing concentration and customer-financing risks that could amplify volatility; management’s statements and backlog additions materially de-risk the $500 billion TAM claim but do not eliminate exposure to hyperscaler spending cycles. Investors should watch incoming order announcements, capex guidance from hyperscalers, and any changes to customer concentration as the primary near-term indicators of sustained upside or renewed downside risk.

Key Decisions for Investors

• Consider increasing selective exposure to NVDA given the Q3 beat, raised guidance, and management's $500 billion 2025–26 sales framework while sizing positions to account for potential volatility from hyperscaler concentration
• Monitor order-backlog disclosures and incremental deal announcements (e.g., Anthropic expansion, Saudi deal) and use them as add-on triggers rather than buying solely on headline optimism
• Avoid or underweight smaller AI-infrastructure names that showed large drawdowns (CoreWeave, Palantir) until their financing and revenue stability are clarified, because customer financing risks are highlighted by management
• Hedge or trim positions if hyperscaler capex guidance softens materially or if Nvidia reports a rising share of revenue tied to a narrower set of customers, since those scenarios would increase downside risk