Export Controls - Page 22

NVIDIA’s LoGRA paper shows low-rank gradient sketches cutting average RL training memory by up to 45.7% on tested Qwen reasoning setups—and making 27B RL runnable on one 8×H100 node where dense Adam OOMs. What the method is, what Table 1 shows, and what it does not prove.

AM Intelligence says two further binding orders for ~20,000 Nvidia Rubin GPUs take committed capacity to ~29,000 GPUs and close to 100 MW, with plans for another 300 MW backed by more than $20bn. Company-reported.

Business Standard relays Bloomberg/FT reporting that SpaceX is in early talks to borrow about $40bn for Nvidia chips. Claims rest on unnamed sources; talks may not close. Not company-confirmed.

Foxconn’s July–September revenue rose 47% to NT$3.03 trillion (about $95.4 billion), beating estimates as AI server demand stayed strong, Reuters reported via Moneycontrol.

Reactor says NVIDIA’s NVentures and Sapphire Ventures joined its Lightspeed-led Series A to scale real-time world-model infrastructure. Secondary outlets report about $74 million in total funding.

TechCrunch, citing anonymous sources, says AI chip startup Etched is reviewing early funding offers at roughly $40–50 billion valuation. Etched declined to comment; the talks are unverified.

Nvidia shares closed at a record $238.90 on 5 October, briefly lifting its market value to about $5.78 trillion, according to Seoul Economic Daily. Here is what is driving the rally and what the numbers do and do not mean.

A Microsoft white paper on "sovereign AI" with Nvidia technologies argues that governments should decide how much control they need workload by workload. It is useful, but it is vendor guidance, not independent policy.

Nvidia and Foxconn reported robot success rates above 95% on GB300 NVL72 tester-tray busbar assembly and 90–95% on multi-connector insertion, according to coverage of Nvidia’s technical blog. Figures are company-reported.

On 7 September 2026 China's Supreme People's Court published Opinions on hearing AI-related disputes: 24 articles covering fault-based liability, notice-and-takedown, training-data evidence, open-source safe harbour and AI-generated court filings. These are court guidance applying existing statutes — China has no dedicated AI law yet.