NVIDIA’s AI Surge Fuels Data‑Center Boom While China Sales Freeze
A wave of AI demand has turned NVIDIA’s data‑center division into a cash‑generating juggernaut, but the same surge comes with an abrupt halt to sales in its biggest overseas market. The quarter’s numbers tell a story of soaring revenue and margins that stay flat only because a $4.5 billion inventory charge vanished – not because costs are under control.
*
When NVIDIA filed its Q1 FY2027 10‑Q on May 20, the headline numbers were spectacular: revenue up 85 % YoY to $81.6 billion and net income more than tripling to $58.3 billion. Yet the real narrative lives in the segment tables.
Data‑center sales exploded. The division posted $75.2 billion of revenue, a 92 % jump from the same quarter a year ago. Hyperscale customers alone contributed $37.9 billion – up from $17.6 billion – while AI Clouds, Industrial & Enterprise added another $37.4 billion versus $21.5 billion last year. Compute & Networking revenue surged 88 % YoY and graphics revenue rose 58 %, both driven by the Blackwell architecture that now dominates the product mix.
Margins stayed flat, but for the wrong reason. Management says gross margin will “remain flat sequentially” because Blackwell continues to dominate revenue. The underlying margin actually improved: a $4.5 billion inventory provision recorded in FY2026 is gone, lifting gross‑margin into the mid‑70 % range. In other words, the quarter’s profitability rests on a one‑time accounting clean‑up rather than operational efficiency.
China sales are effectively shut down. The filing flags that “Data‑center segment sales to China are effectively halted; only uncontrolled products (gaming and workstation GPUs) continue to ship.” Export‑control licensing restrictions, a 25 % import tariff on H200 shipments, and pre‑shipment inspections have turned the Chinese data‑center market from a growth engine into a dead weight. The risk is not just a short‑term dip – it could reshape NVIDIA’s long‑run TAM if China pivots to domestic rivals.
Capital allocation shifts toward lease‑back infrastructure. While the company still holds $50.3 billion in cash and marketable securities, it disclosed a plan to incur $32.4 billion of operating‑lease obligations for data‑center facilities between Q2 FY2027 and FY2033. Lease costs rose to $171 million this quarter (cash outflow $185 million), signaling that NVIDIA is financing the AI boom not by buying more fab capacity but by leasing massive server farms.
The risk profile has expanded. The new 10‑Q adds several risk factors absent from the FY2026 10‑K: a specific focus on Rubin platform delays, heightened exposure to U.S. export controls on H200 products, and an explicit mention of “large, ill‑liquid investments in private companies, infrastructure funds, and publicly‑held equity securities” as a source of volatility. These additions underscore that the AI surge is now being balanced against geopolitical friction and a growing balance‑sheet exposure to non‑core assets.
What it means for investors. The data‑center boom has turned NVIDIA into a de‑facto infrastructure provider, with revenue growth that dwarfs its historic gaming roots. But the China shutdown removes a sizable upside and introduces regulatory headwinds that could push customers toward domestic alternatives. Meanwhile, the massive lease commitment ties future cash flow to the health of the AI compute market – a market that is still subject to energy‑capacity constraints and financing limits for hyperscalers.
Analysts are already debating whether the flat margin guidance masks an underlying earnings quality issue. If the inventory charge disappears permanently, margins could climb into the high‑70s; if new tariffs or export controls bite deeper, the data‑center growth rate could stall, leaving NVIDIA with a costly lease portfolio and limited avenues for offsetting revenue.
The quarter ends with $13.2 billion in cash on hand – down from $62.6 billion at year‑end FY2026 – but bolstered by $50.3 billion of operating cash flow. The liquidity cushion appears sufficient for the next twelve months, yet it is being channeled into lease obligations rather than traditional capex.
Bottom line: NVIDIA’s AI‑driven data‑center surge is real and massive, but it arrives with a geopolitical blind spot in China and a balance‑sheet pivot toward leased infrastructure. The company’s future earnings will hinge on how quickly it can navigate export controls while keeping the AI compute demand engine humming.
--- Key takeaways - Data‑center revenue doubled YoY, pushing total quarterly revenue to $81.6 billion. - Gross margin flatness masks a $4.5 billion inventory charge reversal; true margin pressure remains low. - Sales to China’s data‑center market are effectively halted by export controls and tariffs. - NVIDIA will lock in $32.4 billion of operating‑lease obligations through 2033, shifting capital spending from owned fab capacity to leased AI farms. - New risk disclosures highlight Rubin platform delays, heightened export‑control exposure, and volatility from large illiquid equity investments. ---
Financial Details
| Revenue Guidance | No specific quantitative revenue guidance is provided. Management indicates revenue growth will be driven by data‑center products and AI solutions, but acknowledges potential constraints from energy, data‑center capacity, and customer financing. |
| Capex Plans | Not explicitly disclosed; however, the company plans to incur $32.4 billion of future operating‑lease obligations for data‑center facilities between Q2 FY2027 and FY2033, indicating significant capital commitment to infrastructure. |
| Margin Outlook |
|
| Segment Trends |
|
| Cash Flow Outlook |
|
Key Takeaways
- Data‑center revenue surged 92 % YoY, now accounting for over 90 % of total sales.
- Margins appear flat only because a $4.5 billion inventory provision disappeared; underlying gross margin is improving.
- China data‑center shipments are effectively stopped, exposing NVIDIA to geopolitical and competitive risk.
- The company committed to $32.4 billion of operating‑lease obligations for AI infrastructure through 2033.
- Risk factors now include Rubin platform delays, export‑control licensing constraints, and volatility from large private‑equity holdings.