AI-generated analysis, not investment advice. The articles are written by AI, edited, and checked against company filings — but the judgements are opinions and the figures go stale. How this is made · Terms
Huawei and the Market Nvidia Was Ordered to LeaveThin moat
Nvidia (NVDA) — moat facet
The one competitive loss Nvidia didn't suffer on merit — and the surest way to build a rival ecosystem is a large market where CUDA is banned.
The competitive loss that actually happened was not won on merit. Export controls progressively barred Nvidia's best accelerators from China, and the company's data-center revenue there went to effectively nothing — a market where it had previously held almost the entire AI-silicon share. Huawei moved into the vacuum: its Ascend line has scaled into mass production and it is now credited with roughly half or more of Chinese AI silicon, against a single-digit forecast for Nvidia1.
What makes this more than a lost quarter is what it built. A protected domestic market let Huawei fund several generations of accelerator development it might never have justified commercially, and Chinese frontier labs began optimising their models for Ascend hardware rather than waiting for American chips. Nvidia's moat has always been the software ecosystem rather than the silicon, and the one thing guaranteed to build a rival ecosystem is a large market where CUDA is unavailable.
The uncomfortable arithmetic for shareholders is that this competitor was created by policy and cannot be answered by product. Nvidia can build a faster chip; it cannot sell it there. Some of the business did come back: Hopper shipments to China were $2.4 billion in the third quarter of fiscal 2027, about 2% of data-center revenue, though Nvidia's own fourth-quarter outlook again assumes nothing from the market at all2. Watch whether Huawei's accelerators appear outside China — in the sovereign and emerging-market projects Nvidia is courting — because a domestic champion is a contained problem, and an exporting one is a permanent second ecosystem.
A market Nvidia once owned almost entirely is now largely Huawei's, and the loss compounds: protected demand funded several Ascend generations, and Chinese labs began optimising models for that hardware rather than for CUDA. Nvidia can build a better chip but cannot sell it there, so product cannot answer this. Watch for Ascend appearing outside China.
Some of the business came back after a quarter at less than 1%, and Nvidia's own fourth-quarter outlook again assumes none of it. The number that matters more is whether Huawei's accelerators start appearing outside China.
- Third-party estimateNvidia's Chinese AI-chip share fell from near-total toward a single-digit forecast while Huawei's Ascend took roughly half or more, with Chinese frontier models optimised for Ascend.Market coverage of China AI silicon — NVIDIA's share of the Chinese AI-chip market collapsing from near-total toward a single-digit forecast as export controls took hold, with Huawei's Ascend line scaling into roughly half or more of the market; Chinese frontier models reported as optimised for Ascend hardware — 2026 · publ. 2026 · source ↗
- ReportedHopper shipments to China were $2.4 billion in the third quarter of fiscal 2027, about 2% of data-center revenue, though the fourth-quarter outlook again assumes nothing from the market at all.NVIDIA Q3 FY2027 CFO commentary (Form 8-K, exhibit 99.2) — Data Center revenue of $107.6 billion splits into Hyperscale $60.1 billion (up 71% year on year, up 23% sequentially) and AI Clouds, Industrial & Enterprise $47.5 billion (up 195%, up 18%), growth 'driven by the introduction of our Rubin infrastructure, our fastest architecture ramp ever, standalone Vera CPUs and the continued growth of Blackwell Ultra'; shipments of Data Center Hopper products to China during the quarter were $2.4 billion; Edge Computing $8.2 billion, up 42% and up 14% sequentially, as 'Strong Blackwell workstation sales and edge compute for Physical AI applications drove the increases, while higher memory and system prices continue to affect consumer PC demand'; reportable segments Compute & Networking $106.0 billion (up 108%) and Graphics $9.8 billion (up 62%); GAAP gross margin 74.8% against 75.0% in Q2 FY2027 and 73.4% in Q3 FY2026, higher year on year 'on improved mix from Blackwell Ultra' and down sequentially 'reflecting the initial Rubin ramp'; GAAP operating expenses up 57% year on year and 9% sequentially — Q3 FY2027 (quarter ended 25 October 2026) · publ. 2026-11-17 · source ↗