OTWopentechwire
Tech Intelligence, Openly Wired
Startups

Nvidia Posts $115.8bn Quarter as Leadership Vacuum Tests Partner Ecosystem

Interim CEO Colette Kress delivered results that calmed markets, but the abrupt exit of Jensen Huang has exposed dependencies across the chip maker's web of suppliers, cloud partners and investors.

WZ
Wei Zhang
China Tech Correspondent · Hangzhou
Nov 21, 2026
5 min read
Nvidia Posts $115.8bn Quarter as Leadership Vacuum Tests Partner Ecosystem
Credit: Davide Bonaldo / Dreamstime

A $800bn Question Mark

When Jensen Huang stepped down earlier this month, Nvidia shed roughly $800 billion in market capitalisation within five days. That figure alone dwarfs the combined value of several large Asian tech groups. The 17 November earnings release, the first since his departure, offered partial reassurance: fiscal Q3 2027 sales climbed to $115.8 billion, more than doubling year-on-year, while net income rose 105 per cent to $65.3 billion. Data centre operations generated $107.6 billion, representing 93 per cent of total turnover.

Shares rebounded as much as 4 per cent in extended trading before oscillating through the session. Investors appeared relieved that momentum in AI infrastructure spending had not stalled. Yet the turbulence underscores a deeper concern. For more than thirty years Huang was not only the technical architect but also the relationship broker who tied together manufacturing partners, hyperscale customers, and regulators across continents. His abrupt resignation leaves those connections without their principal node.

Colette Kress, who continues to hold the chief financial officer role while serving as interim CEO, now steers a company whose product pipeline remains strong but whose institutional memory and diplomatic capital walked out the door. The board has made clear it will not entertain Huang's return, despite public appeals and an online petition that has drawn more than 180,000 signatures.

China Revenue Surges, Yet Access Remains Constrained

Revenue attributed to customers based in mainland China and Hong Kong reached $11.05 billion in the quarter, up approximately 272 per cent. That geographic label, however, measures where direct buyers are headquartered rather than the ultimate destination or use of the silicon. The company's guidance for the current quarter assumes zero sales of data centre computing products into China, a reflection of export restrictions that have tightened over the past year.

On 16 November the US Department of Commerce introduced rules targeting arrangements that let restricted buyers tap advanced American chip capacity through data centres in third countries. The measures extend beyond physical hardware shipments to remote inference and training workloads. Chinese cloud providers and AI model developers had been booking overseas GPU capacity in advance, hoping to lock in supply before further controls took effect. Those reservations now carry less certainty.

A salesperson at one cloud provider told Chinese technology outlet 36Kr that tighter supply could actually improve margins for operators with existing inventories. Demand for training and inference continues to climb, and if new high-end GPU shipments slow, companies holding capacity stand to gain pricing power. The gap between firms that secured hardware early and those still hunting for allocations is likely to widen.

Domestic Alternatives Gain Ground

Before Huang's exit, state-owned enterprises in China were the main organisations piloting local accelerators from vendors such as Huawei, Cambricon and Moore Threads. Startups hesitated because adapting software stacks required engineering time they could not spare while racing to ship products and close funding rounds. Unpredictable access to foreign chips is now forcing a reassessment.

In February, Z.ai's GLM-5 model added compatibility with several Chinese platforms, including Huawei Ascend. Around the same time, Tecorigin announced interoperability with more than thirty foundation models. By September, Ascend had deepened its integration with PyTorch, which now lists the platform among its supported backends. Progress is moving from one-off model ports to broader infrastructure layers.

Local subsidy programmes are reinforcing the shift. One industry participant said that in recent months certain projects have begun requiring at least 80 per cent of accelerator capacity to come from domestic suppliers. An executive at a cloud company described preparing a contingency plan centred on Chinese hardware, acknowledging short-term friction but expecting medium-term benefits for the sector's self-sufficiency.

A Web of Investments Under Stress

Over the past year Nvidia expanded its footprint across the AI supply chain through equity stakes, purchase commitments and technology partnerships. It invested in Intel, Synopsys and Marvell; committed billions to optical component makers Lumentum and Coherent; and struck capital and technology agreements with Corning and Nokia. It also backed cloud infrastructure operators CoreWeave and Nebius, which in turn buy and deploy Nvidia silicon at scale.

The structure places Nvidia simultaneously as supplier, investor, customer and financier. By securing upstream manufacturing capacity while funding downstream demand, the company drew more participants into its orbit. That architecture proved effective during the build-out phase of generative AI infrastructure. It also created channels through which uncertainty at the centre can radiate outward.

Between 11 and 16 November, shares of CoreWeave, Nebius, Lumentum, Coherent, Corning, Intel, Marvell, Nokia and Synopsys all recorded steep declines. After Nvidia's results landed, CoreWeave and Nebius climbed roughly 5 and 6 per cent respectively in after-hours trading, while Intel and other semiconductor stocks also recovered ground. The swift reversal illustrates how tightly investor confidence in Nvidia now correlates with sentiment toward its extended network.

Board Formalises Geopolitical Oversight

Ahead of the earnings release, Nvidia's board established a standing committee to oversee national security policy, export compliance and global trade risks. The move introduces formal governance in areas where Huang's personal relationships with officials in Washington and partner capitals previously carried significant weight.

Chairman Harvey Jones opened the earnings call by ruling out any prospect of Huang's reinstatement and declining to take questions on his departure. In a public statement on 17 November, Huang acknowledged "serious mistakes in my personal life" and said he had created problems for the board that he deeply regretted. He added that if directors decided his return served shareholders, employees and the country, he would be honoured to resume the role.

Outside Nvidia's headquarters in Santa Clara, white flowers left by supporters remained in place days after his resignation. The Financial Times reported that allies were organising a "war room" of Silicon Valley executives to lobby the White House for his reinstatement. The effort reflects the loyalty Huang commands among some industry figures, yet the political and commercial landscape has shifted in ways that may limit any individual's ability to navigate it as he once did.

What Continuity Looks Like

Nvidia's quarterly performance demonstrates that product development and customer demand have not faltered in the immediate wake of leadership change. Data centre revenue grew 110 per cent, and the pipeline for next-generation architectures remains on schedule. Kress has been with the company for more than a decade and understands its financial and operational rhythms.

The harder question is whether the pace of investment, the density of partnerships, and the diplomatic finesse that characterised Huang's tenure can be replicated by committee and process. The new geopolitical oversight body signals an attempt to institutionalise what was previously personal. Success will depend on whether formal structures can substitute for the accumulated trust and technical credibility that one individual brought to negotiations with suppliers, customers and governments.

Chinese companies are accelerating their shift toward domestic silicon, not out of preference but necessity. American partners are recalibrating their exposure to a supplier whose leadership continuity is now in doubt. Nvidia's quarterly sales offer a snapshot of the business Huang built. The test for his successors is whether that business can sustain its trajectory when the relationships that underpinned it are no longer guaranteed.

Read next
Startups

NTT Bets on Asset-Light Data Centre Model to Cut Debt and Fund Expansion

Kenji Watanabe · 6 min
Startups

Anthropic Debuts at $1.5 Trillion Valuation as IPO Exceeds Target Price

Arjun S. Mehta · 6 min
Startups

Bengaluru's Humyn Labs Secures $47 Million to Supply Physical AI Training Data

Hana Park · 7 min
Spot something wrong? Email corrections@opentechwire.com. We log every correction publicly.