NVIDIA

words.The S&P 500 slipped after a fresh high, driven by a tech‑stock rotation while materials, financials and industrials led gains; the Dow rose 1 %. Investors await the “Santa Claus rally” starting Dec 19. Key week‑long stories: Broadcom fell 11.5% on cautious AI‑chip demand; Oracle dropped further after delaying OpenAI data‑center projects; Nvidia secured limited export licences for a throttled AI accelerator to China; GE Vernova posted strong guidance on AI‑data‑center power‑equipment. Market focus now is Fed policy, AI‑chip supply‑chain dynamics, and enterprise‑software spending.

Nvidia denied reports that Chinese AI startup DeepSeek smuggled its advanced Blackwell chips, which the U.S. has banned from export to protect a strategic AI lead. While Nvidia investigates all leads, it says there’s no evidence of “phantom data centers.” President Trump has proposed allowing H200 chips to “approved” Chinese customers with a 25% U.S. revenue share, sparking bipartisan debate over technology‑transfer risks. DeepSeek’s low‑cost R1 model has risen quickly, highlighting competitive pressure on Western AI firms and the uncertainty of export controls.

.Nvidia received U.S. approval to export its second‑tier H200 GPUs to vetted Chinese customers for a 25 % treasury royalty, a move announced by President Trump. The H200, an upgraded version of the throttled H20, could add $25‑30 billion in revenue and lift EPS by $0.60‑$0.70, yielding a PEG below 1.0 and a forward P/E around 24×. Though Wall Street’s reaction was muted, analysts expect Chinese adoption to narrow the AI compute gap and boost Nvidia’s earnings despite lingering geopolitical risks.

U.S. authorities dismantled a China‑linked smuggling ring that moved over $160 million of export‑controlled NVIDIA H100 and H200 GPUs. Operation Gatekeeper led to the guilty plea of Texas businessman Alan Hao Hsu and his firm Hao Global, plus charges against New York‑based Fanyue Gong and Canadian Benlin Yuan for using falsified documents and “straw purchasers” to reroute chips to mainland China, Hong Kong and other prohibited sites. The bust underscores heightened enforcement of AI‑hardware export controls and the broader U.S.–China rivalry over advanced semiconductor technology.

President Trump announced that Nvidia may ship its H200 AI chips to approved Chinese customers, with the United States receiving 25% of sales revenue. Chinese President Xi Jinping welcomed the plan. The same revenue‑sharing model will apply to AMD, Intel and other U.S. chipmakers. The policy is presented as a way to protect American jobs, sustain domestic manufacturing and generate tax income while avoiding a full export ban. Analysts see modest share gains, increased competition, and a shift toward calibrated engagement that could become a template for future high‑tech export agreements.

the summary).Foxconn, traditionally a smartphone assembler, is now a key Nvidia partner, producing AI‑optimized server racks that drove a 26% YoY revenue rise to NT$844.3 bn (≈US$27 bn) in November. AI server sales offset a 6% monthly dip in consumer electronics, boosting Q3 profit 17% YoY. The firm launched a Taiwan AI manufacturing hub with Nvidia, invested in data‑center specialist TECO, and partnered with OpenAI on hardware. This pivot promises higher margins, greater capital spend, and geographic diversification, lifting its share price 26% YTD and positioning Foxconn as a central player in the expanding AI infrastructure market.

. Nvidia announced this week that it will invest $2 billion for a stake in chip‑design fir…

Nvidia CEO Jensen Huang met former President Donald Trump on Capitol Hill to discuss the proposed GAIN AI Act, which would force U.S. chipmakers to prioritize domestic sales of advanced AI processors over exports to markets such as China. Huang welcomed the bill’s exclusion from the NDAA, warning it would stifle innovation and harm U.S. competitiveness, and advocated for a single federal AI framework rather than fragmented state rules. He noted that tighter export controls could reshape global AI‑hardware supply chains, prompting firms like Nvidia and AMD to shift focus toward domestic and allied‑nation customers while navigating stricter compliance requirements.

words.Anthropic, the AI startup behind Claude, is preparing for a potential IPO—the largest slated for next year—by hiring law firm Wilson Sonsini and consulting banks. The company may raise a private round valued over $300 billion, with $15 billion pledged by Microsoft and Nvidia. It has hired former Airbnb IPO lead Krishna Rao and announced a $50 billion data‑center expansion. Rival OpenAI is also weighing a public listing, positioning both firms to test market appetite for fast‑growing, loss‑making AI ventures.

summary.Alphabet’s Google has surged ahead of OpenAI after releasing Gemini 3, a large‑language model that matches or surpasses ChatGPT, and unveiling its seventh‑gen “Ironwood” TPU. Both Google and silicon partner Broadcom posted strong quarterly gains, while OpenAI‑backed Nvidia and Microsoft lagged. Analysts note a pricing premium for Google’s AI assets for the first time in a decade. The move raises questions about TPU third‑party revenue, Nvidia’s GPU moat, and whether Google can sustain AI‑driven profit growth beyond advertising.