NVIDIA CFO: $100 Billion OpenAI Deal Not Finalized; Lead “Has Absolutely Not Narrowed”

NVIDIA OpenAI $100 billion deal is not finalized, CFO says — the letter of intent remains non‑binding and NVIDIA’s lead “has absolutely not narrowed.”

Key Points

Non‑binding LOI: Reported $100 billion OpenAI arrangement (≈ ¥720 billion) to deploy at least 10 gigawatts of systems — the agreement is a non‑binding letter of intent and any revenue is not included in current guidance.
Potential scale if realized: Media estimates suggest roughly $400–$500 billion in long‑term revenue (≈ ¥2.88–¥3.60 trillion), but this is contingent until contracts are signed and revenue is recognized.
Competitive and product position: Yīngwěidá 英伟达 stresses its lead “absolutely not narrowed,” highlighting a system‑level advantage, the upcoming Vera Rubin architecture (ship H2 next year), and expected gross margins around the mid‑70% range.
Operational signals to watch: NVIDIA saw about a $25 billion rise in combined inventory and purchase commitments, is prioritizing supply expansion (then shareholder returns), and has prior commitments like up to $10 billion for Anthropic.

Overview — NVIDIA OpenAI $100 billion deal update

At the UBS Global Technology and Artificial Intelligence Conference, NVIDIA (Yīngwěidá 英伟达) Chief Financial Officer Colette Kress spoke directly about headlines describing a reported “hundreds‑of‑billions” cloud deployment for OpenAI.

She confirmed the September announcement remains a non‑binding letter of intent and is not a signed, final contract.

She also pushed back on any suggestion that NVIDIA’s competitive advantage is slipping, saying the company’s lead “absolutely not narrowed.”

Resume Captain

Your AI Career Toolkit:

AI Resume Optimization
Custom Cover Letters
LinkedIn Profile Boost
Interview Question Prep
Salary Negotiation Agent

Get Started Free

Key takeaway on the OpenAI arrangement

Kress reconfirmed the reported arrangement to deploy “at least 10 gigawatts” of NVIDIA systems for OpenAI is still at the letter‑of‑intent stage.

The deal has been widely reported as on the order of $100 billion, which equals approximately ¥720 billion RMB using the exchange rate cited below.

Kress emphasized that any revenue from that cooperation is not included in NVIDIA’s current guidance.

What the reported scale would mean for NVIDIA — context and implications

Externally, the announcement has been framed as potentially generating roughly $400–$500 billion in revenue for NVIDIA over time if full deployment materializes.

That range converts to approximately ¥2.88–¥3.60 trillion RMB at the conversion used in the talk.

But Kress was clear: no final agreements have been signed, and NVIDIA is still working with the parties involved.

Why that matters for investors and operators:

Forecast variability: Headline numbers can be transformative if realized, but they remain contingent until contracts are signed and revenue is recognized.
Supply and capacity planning: NVIDIA is already prioritizing capital to expand supply and ramp next‑gen products like Vera Rubin, reflecting how the company prepares for large, conditional orders.
Market signaling: Even non‑binding LOIs can shape competitors’, partners’, and suppliers’ behavior quickly — from component vendors to cloud providers.

Find Top Talent on China's Leading Networks

Post Across China's Job Sites from $299 / role, or
Hire Our Recruiting Pros from $799 / role

- - - - - - - -

Qualified Candidate Bundles
Lower Hiring Costs by 80%+
Expert Team Since 2014

Get 25% Off
Your First Job Post

Other notable points from the interview — quick bullets investors and founders should care about

Deal pipeline and Anthropic: NVIDIA previously pledged up to $10 billion to Anthropic, which could increase chip reservation volumes and pre‑orders. ($10,000,000,000 USD ≈ ¥72,000,000,000 RMB)
AI industry shift: Kress described a structural move where many data‑center workloads that ran on CPUs are migrating to GPUs, and she estimated $3–4 trillion will be invested in data‑center infrastructure this decade, with roughly half tied to that CPU→accelerated computing transition.
System‑level lead: NVIDIA’s advantage is not just one chip; it’s a system‑level co‑design across chips plus a full hardware‑and‑software stack — especially CUDA and extensive libraries — which keeps most models running on NVIDIA platforms across cloud and on‑premises.
Customer economics and flywheel: As workloads shift from training to inference and end users pay for tokens/services, expanding revenue can enable larger models and more compute investment, improving customer economics over time.
Order risk and pre‑bookings: NVIDIA evaluates purchase orders (POs) and customer funding capability, since many model developers place large compute pre‑orders but may have limited near‑term revenue.
Vera Rubin (next‑gen architecture): The Vera Rubin architecture has completed tape‑out and is expected to ship in the second half of next year with multiple‑times performance improvements versus prior generations.
Margins: Despite rising HBM memory costs, NVIDIA expects to maintain gross margins around the mid‑70% range next year.
Inventory and purchase commitments: Last quarter NVIDIA saw about $25 billion of combined inventory and purchase commitments increase, a large jump from prior quarters’ increases in the single‑digit‑hundreds‑of‑millions, indicating supply buildup for the next growth wave.
Capital allocation: NVIDIA’s guidance points to revenue next year in the $350–$400 billion range (¥2.52–¥2.88 trillion RMB), and the company plans to prioritize supply capacity expansion, then shareholder returns and ecosystem investments, while favoring engineering‑led acquisitions over large M&A.

Numbers called out in the talk (with currency conversions)

Reported OpenAI deployment estimate: $100 billion (≈ ¥720 billion RMB)
Potential revenue impact often cited for the OpenAI deployment: $400–$500 billion (≈ ¥2.88–¥3.60 trillion RMB)
Committed potential investment in Anthropic: up to $10 billion (≈ ¥72 billion RMB)
NVIDIA guidance cited for next year revenue: $350–$400 billion (≈ ¥2.52–¥2.88 trillion RMB)

Conversion used for the figures above: 1 USD ≈ ¥7.20 CNY (rates fluctuate; conversions are approximate and rounded for clarity).

Investor and founder takeaways — what to watch next

Track whether conditional LOIs convert to signed contracts and when revenue recognition starts.

Watch Vera Rubin shipping timing and early performance data as a proxy for NVIDIA’s product roadmap execution.

Monitor NVIDIA’s inventory flow — a big jump in inventory and purchase commitments can indicate supply pre‑positioning for large customers.

Follow the capital allocation cues: supply expansion first, shareholder returns second, selective ecosystem investments after that.

For startups and cloud providers building on NVIDIA stacks, CUDA and NVIDIA libraries remain core due to the installed base and systemized tooling advantage.

Why this matters — short and sharp

Kress’ comments reiterate two central points for investors and customers.

First, transformational headline deals can remain conditional until final agreements are signed and revenue is booked.

Second, NVIDIA’s competitive edge is a system‑level advantage across hardware, software, and ecosystem, which keeps it central to model development and deployment across cloud and on‑premises environments.

The company is allocating capital to secure supply, ramp next‑gen products, and return cash to shareholders while selectively investing in partners that strengthen the platform and R&D.