Key Highlights

  • Nvidia and Safe Superintelligence confirmed a partnership at last closing; Nvidia did not disclose the investment size, and Bloomberg's reported $5 billion figure remains unconfirmed by either party.
  • First-quarter fiscal 2027 revenue reached $81.6 billion, up 85% year over year, with data-centre revenue of $75.2 billion and non-GAAP gross margin of 75.0%.
  • Nvidia guided second-quarter revenue to $91.0 billion while assuming zero data-centre compute revenue from China, a reminder that the growth is arriving despite, not because of, the export-control picture.
  • The SSI partnership is confirmed, the amount is not, and Vera Rubin has not shipped to anyone.

Nvidia (NASDAQ:NVDA) confirmed a long-term strategic partnership and investment in Safe Superintelligence at last closing without disclosing the amount. The more consequential detail is what SSI receives: an order-of-magnitude compute increase on a Vera Rubin platform that has not yet shipped.

A Research Lab With No Product Just Became Nvidia's Newest Partner

Nvidia Corporation (NASDAQ: NVDA) has spent two years converting an accelerator monopoly into an ecosystem franchise; the 27 July partnership announcement with Ilya Sutskever's Safe Superintelligence Inc. is the sharpest expression of that strategy yet. SSI has no product, no disclosed revenue and a stated policy of avoiding commercial releases entirely. What it has is Sutskever, OpenAI's former chief scientist, and, now, a claim on Nvidia's next-generation silicon.

The announcement landed the same day reports surfaced that Nvidia was weighing far larger financing commitments to OpenAI. Investors must hold two ideas at once: demand for Nvidia compute is extraordinary, and Nvidia is increasingly financing the entities generating it.

What Nvidia Actually Confirmed, and What It Did Not

Nvidia's own investor-relations release confirms three things: a long-term strategic partnership with SSI, an equity investment of unstated size, and SSI access to the Vera Rubin platform sufficient to raise its compute capability "by an order of magnitude." Jensen Huang's quoted comment, that Nvidia is excited to see what breakthroughs SSI discovers on Vera Rubin, is forward-looking, an expression of expectation rather than a claim of results.

Bloomberg reported a $5 billion investment; TechCrunch, citing a person familiar with the deal, described it only as "multiple billions." Neither party has confirmed a figure. SSI's $32 billion post-money valuation comes from PitchBook and reflects prior rounds totalling roughly $7 billion from backers including Andreessen Horowitz, Alphabet, Lightspeed and Sequoia. Treat the headline number as reported, not established.

The separate OpenAI arrangements are even less settled. Axios reported Nvidia was exploring a roughly $250 billion backstop tied to an Ohio data-centre project and about $350 billion in chip-purchase financing; representatives for both companies declined to comment. Credit-default swaps on Nvidia bonds recorded their sharpest intraday move since November on the reports, the credit market's way of registering that the buyer and the lender are becoming the same party.

The Quarter Behind the Confidence

Nvidia's most recently reported period is the first quarter of fiscal 2027. (Nvidia's fiscal 2026 closed in January 2026, hence fiscal 2027 now.)

Revenue was $81.6 billion, up 85% year over year and 20% sequentially. Data-centre revenue reached $75.2 billion, up 92%, split between $60.4 billion of compute and $14.8 billion of networking, the latter up 199% year over year, an underappreciated line given that InfiniBand, Spectrum-X Ethernet and NVLink are what make rack-scale systems saleable as systems rather than chips. Edge computing contributed $6.4 billion.

Margins were flat sequentially at 74.9% GAAP and 75.0% non-GAAP. The year-over-year GAAP expansion of 14.4 points is flattered by a base effect: the prior-year quarter absorbed a $4.5 billion charge on excess H20 inventory.

GAAP net income of $58.3 billion exceeded GAAP operating income of $53.5 billion, because gains on Nvidia's investment portfolio flow below the operating line. Non-GAAP net income was $45.5 billion, and non-GAAP diluted EPS was $1.87 against GAAP EPS of $2.39, a rare case where the GAAP number is the more generous one, precisely because of the ecosystem investments now drawing scrutiny.

Cash conversion was strong: $50.3 billion of operating cash flow and $48.6 billion of free cash flow on just $1.8 billion of capital expenditure.

Selling Systems, Buying Demand

Nvidia's commercial model has shifted from selling accelerators to selling AI factories, and Vera Rubin is the vehicle. Nvidia announced that the platform was ramping into full production, integrating Vera Rubin NVL72 systems with the Vera CPU, BlueField storage and Spectrum Ethernet, and claiming roughly 10x agent throughput versus the Grace Blackwell generation. Shipments were stated to begin in autumn 2026, with system builders including Dell, HPE, Lenovo, Supermicro and Foxconn, and cloud partners including CoreWeave, Lambda and Oracle Cloud Infrastructure.

That timeline is the crux of the SSI story. Huang told a Tokyo developer event on 15 July that the platform was "already in production" with "giant amounts of production incoming," pushing back on reports of manufacturing delays; a KeyBanc analyst had flagged an unconfirmed high-bandwidth-memory qualification risk. But production is not the same as validated, deployed racks, and Huang offered no customer-delivery date. Any suggestion that Vera Rubin is presently powering a superintelligence breakthrough at SSI is unsupported, the hardware has not shipped.

Share Is Still Overwhelming, but the Slope Has Changed

Nvidia's estimated share of AI accelerators sits near 70%, with erosion coming from custom silicon, not a rival GPU. Broadcom reported $8.4 billion of AI semiconductor revenue in its fiscal first quarter of 2026, up 106%, with management guiding to more than $100 billion of AI chip revenue in 2027 against a disclosed $73 billion backlog. Google's seventh-generation Ironwood TPU is generally available, with Anthropic committed to as many as one million TPUs; AWS says it has deployed more than a million Trainium processors, and Microsoft's Maia 200 entered deployment in January 2026.

AMD is the credible merchant alternative: Helios racks and the MI400/MI450 family reached production in 2026, underpinned by 6-gigawatt commitments from OpenAI and, separately, Meta. The realistic reading is not displacement but compression, Nvidia's customers are cultivating second sources, which constrains pricing power more than volume.

A Fortress Balance Sheet Being Pointed Outward

At Nvidia held $13.2 billion of cash and equivalents plus $67.3 billion of marketable securities, of which $30.2 billion sat in equity securities. Debt was modest: $1.0 billion short-term and $7.5 billion long-term against $195.5 billion of shareholders' equity. The company returned approximately $20.0 billion to shareholders in the quarter, added $80.0 billion to its repurchase authorisation and raised the quarterly dividend from $0.01 to $0.25 per share.

Two commitments define the risk profile. Inventory rose to $25.8 billion and total supply-related commitments stood at $119.0 billion, management has pre-bought capacity well beyond near-term needs. The equity portfolio, which includes a reported $30 billion stake in OpenAI and a $10 billion commitment to Anthropic, means Nvidia's earnings now carry mark-to-market exposure to the valuations of its own customers.

Where the Next Leg Would Come From

The nearest catalyst is second-quarter guidance of $91.0 billion in revenue, plus or minus 2%, with margins near 75%, a sequential increase of roughly $9 billion achieved while assuming nothing from China. Vera Rubin volume shipments in the autumn are the second. A third is optionality rather than forecast: the Commerce Department revised its licence-review policy to permit H200-class exports to China case-by-case under security conditions.

What Could Break

China remains unresolved. No Hopper data-centre products shipped there in the quarter, versus $4.6 billion a year earlier, and US authorities said in June that restrictions extend to Chinese firms operating outside China. Concentration is real: hyperscalers accounted for roughly 50% of data-centre revenue. The $119.0 billion of supply commitments would become a costly liability if orders soften, the prior year's H20 write-down shows how quickly that arithmetic turns. The circular-financing critique is now a credit-market concern, not just a rhetorical one.

What the Price Implies

Nvidia closed at $196.51 at last closing, for a market capitalisation of about $4.76 trillion and enterprise value near $4.72 trillion, on a trailing P/E of roughly 30, a forward P/E of about 20 and a price-to-sales ratio near 18.8, with trailing free cash flow of $119.1 billion (based on aggregated market data; multiples move continuously). The forward multiple is the striking figure, it sits below the broad-market average and reflects consensus expectations of continued rapid earnings growth. That is the bet embedded in the price: not that Nvidia is cheap on earnings, but that its growth estimates are achievable.

Conclusion

The SSI partnership is confirmed, the amount is not, and Vera Rubin has not shipped to anyone. Those three facts are the story. Nvidia is using a balance sheet built on 75% gross margins to seed the research labs that will consume its next platform, $48.6 billion of quarterly free cash flow is not a narrative. The open question is whether an ecosystem increasingly capitalised by its principal supplier can absorb the compute it has committed to buy. The SSI announcement will not settle that, and probably neither will the August quarter.

This article is for informational purposes only and does not constitute investment advice or a recommendation to buy, sell, or hold any security. Share prices can rise or fall, and past performance does not guarantee future results. Investors should conduct independent research and consult a licensed financial adviser before making investment decisions.