Nvidia’s AI Infrastructure Push Faces Hurdles as OpenAI Deal Remains Unsigned

Nvidia’s data center business continues to surge, fueled by the explosive growth of artificial intelligence, but key partnerships and market access remain points of uncertainty for the chipmaker. While recent earnings exceeded expectations, questions linger around a massive investment deal with OpenAI and ongoing restrictions on sales to China.

Data Center Revenue Soars, Driven by AI Demand

Nvidia’s data center unit generated $19.37 billion in revenue, a 68% increase year-over-year, accounting for over 91% of the company’s total sales. This growth has been substantial, with the data center business scaling nearly 13x since the emergence of ChatGPT in fiscal year 2023, according to Nvidia CFO Colette Kress . This surge aligns with projections of a $2.5 trillion global AI spending market this year and significant capital expenditure commitments from major hyperscalers like Amazon, Alphabet, Meta, and Microsoft .

$100 Billion OpenAI Deal Still Pending

Despite announcing a plan in September 2025 to deploy at least 10 gigawatts of Nvidia systems and invest up to $100 billion in OpenAI, the agreement remains unfinalized. Colette Kress confirmed that a “definitive agreement” has not yet been completed , . This clarification casts doubt on the initial perception of the deal, once touted by Nvidia CEO Jensen Huang as “the biggest AI infrastructure project in history” . Nvidia’s latest filings indicate “no assurance that any investment will be completed on expected terms, if at all” , extending to planned investments in Anthropic and Intel as well.

Hyperscaler Capex and the Question of ROI

Major hyperscalers have committed nearly $700 billion in capital expenditure for 2026, raising concerns among investors about potential cash flow issues . Despite these substantial investments in AI infrastructure, the economic impact remains unclear. A recent Goldman Sachs analysis suggested that AI contributed “basically zero” to U.S. GDP in 2025 .

The Need for Enterprise AI Adoption

Huang emphasized the importance of “agentic AI” and its potential to drive cash flow growth for hyperscalers. However, broader enterprise adoption of AI is crucial to alleviate concerns about an AI bubble. Recent surveys indicate that despite 70% of firms employing AI, over 80% have reported no impact on employment or productivity . Huang highlighted Anthropic’s Claude Cowork as a potential catalyst for increased AI penetration in the workforce.

Compute Power as the New Revenue Driver

Huang argued that compute power—specifically, the generation of tokens (chunks of data processed by AI models)—is directly correlated with revenue. As AI models become more complex, the demand for computing power increases “exponentially,” justifying the large capital expenditures . He believes that the continued build-out of data center capacity will power higher-level models and ultimately translate to revenue.

China Market Access Remains Uncertain

Nvidia has recently been permitted to ship small amounts of its H200 chips to China, following previous restrictions imposed by the Trump administration . However, the long-term stability of this access remains uncertain, and the company is not currently factoring it into revenue projections.

Colette Kress has served as NVIDIA’s executive vice president and chief financial officer since September 2013 .