Nvidia’s AI Guarantee Exposure Reaches $118 Billion

TSMC remains a central supplier to the AI boom, with roughly 73% of the global chip foundry market and about 90% of the most advanced processor market. The articles cite third-quarter sales growth of about 41% to roughly $46.3 billion–$46.72 billion and a sixth consecutive quarter of record profits. TSMC expects fourth-quarter revenue of $52.3 billion to $54.3 billion and forecasts AI accelerator revenue growth above 50% annually through 2029. October revenue rose 59.7% year over year, while revenue for the first ten months of 2026 climbed 43.3%. Fisher Asset Management held about 18.93 million TSMC shares, worth $8.64 billion, at September 30. Risks include Taiwan production concentration, the expense of building fabs elsewhere, and the possibility of slower AI orders.

One comparison casts ASML and TSMC as longer-term alternatives to Nvidia after Nvidia shares fell 13% in a week; ASML and TSMC shares were down 26% and 16% over three months. ASML’s quarterly revenue rose 25% to 11.66 billion euros, and demand for 2027–28 exceeds its supply capacity. The article gives forward price-to-earnings ratios of 16.4 for Nvidia, 20 for TSMC and 29 for ASML. Separately, Nvidia’s potential exposure from AI-infrastructure guarantees rose $9.5 billion in a quarter to $118 billion. Of that, $105 billion backs SB Energy’s Ohio campus, expected to host Nvidia systems leased to OpenAI. Guarantees apply only after buildings are completed, with the first expected to qualify in fiscal 2029, and decline over the 20-year leases as OpenAI pays. Nvidia also reported $32 billion in future neocloud commitments, separate from the guarantees.

Questions about who benefits from AI extend beyond chips. At a November 16 Federal Reserve task force conference, economists said AI’s effect on overall productivity remains small so far, despite workplace uses such as meeting summaries and radiology assistance. Past technology gains have taken years or decades to translate into higher pay, and compensation has recently grown more slowly than productivity. Moody’s Analytics economist Mark Zandi estimates AI has driven nearly two-thirds of U.S. investment growth since ChatGPT launched four years ago, while investment growth elsewhere has slowed by more than half since mid-2024. He says the boom supports growth and job creation but can raise electricity and electronics prices, push up interest rates, and divert investment; it must deliver productivity gains as investment elsewhere weakens.

Big Tech’s ability to withstand AI competition is also in focus. Weakness in some core businesses at Alphabet and Amazon has prompted concern that AI could challenge established companies, even as large technology firms retain strong cash flow, margins and access to low-cost debt. Supporters argue those resources—and investments in firms such as Anthropic—could help them absorb an AI downturn, though investor confidence could still shift away from hyperscalers. In enterprise security, Spin.AI made extension risk assessments available through Microsoft Edge for Business on November 19, 2026. Its analysis covers more than one million extensions and apps and checks over 25 risk factors. Spin.AI says half of extensions in use are high risk; it also cites an average of 1,454 extensions at organizations with over 2,000 employees and the 2025 RedDirection campaign, which affected more than 16 million users.

Several smaller products and research efforts target practical AI use. Wilgot raised a $3 million seed round for tools that use retailers’ searches, reviews, support tickets and returns to improve product pages and structured data for AI shopping. It reports 20% higher net sales across optimized Daniel Wellington pages and says 7% of Myrqvist customers arrive through ChatGPT; it gave no measurement periods or experimental details and cannot control independent assistants’ recommendations. Fensory connects AI trading agents to Hyperliquid and lets users restrict market access and set order-size limits across crypto, stock-linked markets, gold, oil and forex, but those controls do not remove trading and leverage risks. A prompt-optimization study found its simple DCPS method matched StablePrompt’s classification accuracy within 0.5 percentage points at about one-third the measured cost, and matched GEPA’s macro accuracy on GPT-4.1-mini with 55% fewer rollouts and 32% lower average API costs. More complex methods still helped on some open-ended and reasoning-heavy tasks.

Other research and international agreements address adaptation and digital infrastructure. PACT adapts pretrained time-series models to changing conditions by making localized embedding adjustments while preserving earlier learning; it uses 5% labeled calibration data per new condition and does not require the condition to be known at test time. In a four-condition bearing-fault test, average macro-F1 stayed above 0.95, with similar results on public benchmarks involving matching task and label types. Gabon signed three memorandums of understanding with Huawei during President Brice Clotaire Oligui Nguema’s Shanghai visit, covering intelligent public security, public-university digital modernization, and AI teaching equipment for Brice Clotaire Oligui Nguema University.

Key Takeaways

  • TSMC reports about 73% of foundry capacity and 90% of the most advanced processor market; it forecasts AI accelerator revenue growth above 50% annually through 2029.
  • Fisher Asset Management held an $8.64 billion TSMC stake, or 2.48% of its portfolio, at September 30, 2026; Taiwan concentration and slower AI demand remain cited risks.
  • ASML’s latest quarterly revenue rose 25% to 11.66 billion euros, with 2027–28 demand exceeding supply; the article compares forward P/E ratios of 29 for ASML, 20 for TSMC and 16.4 for Nvidia.
  • Nvidia’s potential AI-infrastructure guarantee exposure reached $118 billion, including $105 billion tied to an Ohio campus expected to host Nvidia systems leased to OpenAI; separate neocloud commitments total $32 billion.
  • Federal Reserve economists say AI’s overall productivity impact remains small so far; Mark Zandi estimates AI has driven nearly two-thirds of U.S. investment growth since ChatGPT launched.
  • Alphabet and Amazon earnings showed weakness in some core businesses, renewing investor concerns about AI competition; strong cash flow and investments including Anthropic are cited as potential buffers.
  • Spin.AI’s assessments are available in Microsoft Edge for Business and evaluate more than one million extensions and apps against over 25 risk factors.
  • Wilgot raised $3 million to optimize retail product information for AI shopping and reports higher sales on Daniel Wellington pages and ChatGPT referrals for Myrqvist, without providing measurement details.
  • Fensory lets users constrain AI trading agents’ market access and order sizes on Hyperliquid, but warns that automated trading, leverage and losses remain risks.
  • DCPS matched selected prompt-optimization results at lower measured costs; PACT used 5% calibration data and maintained average macro-F1 above 0.95 in a four-condition bearing test. Gabon also signed three AI, security and education agreements with Huawei.

TSMC stands out as a leading AI chip stock through 2030

Taiwan Semiconductor Manufacturing (TSMC) is presented as a strong choice for investors seeking long-term exposure to AI chip demand. The company holds about 73% of the global chip foundry market and an estimated 90% share of the most advanced processor market. Third-quarter revenue rose about 41% to $46.72 billion, and profit reached a sixth straight quarterly record. TSMC expects fourth-quarter revenue of $52.3 billion to $54.3 billion and says AI accelerator revenue could grow at a high-50% annual rate through 2029.

TSMC grows sales 41% and holds a 90% share of advanced AI processors

Taiwan Semiconductor Manufacturing Company reported third-quarter sales of $46.3 billion, up 41%, and recorded its sixth straight quarter of record profits. TSMC holds 73% of the global chip foundry market and 90% of the market for the most advanced AI processors. The company expects AI accelerator revenue to grow at an annual rate of more than 50% through 2029 as demand for AI infrastructure rises. The article says TSMC’s stock remains fairly valued despite recent market volatility.

ASML and TSMC offer AI exposure as Nvidia shares fall

Nvidia’s stock fell 13% in a week, while ASML and Taiwan Semiconductor Manufacturing shares dropped 26% and 16% over the past three months. ASML makes complex laser machines needed to produce advanced chips, and no other company can match its capabilities. Its latest quarterly revenue rose 25% to 11.66 billion euros, while customer demand for 2027 and 2028 exceeds its current supply capacity. TSMC makes Nvidia’s chips and other advanced processors, reported 60% year-over-year revenue growth in October, and has factories in Taiwan as well as new hubs being built in Japan and the United States. The article cites forward price-to-earnings ratios of 16.4 for Nvidia, 20 for TSMC, and 29 for ASML, and argues that ASML and TSMC are longer-term alternatives to buying Nvidia shares.

Ken Fisher holds an $8.64 billion stake in TSMC

Fisher Asset Management reported a portfolio worth about $348.3 billion for the quarter ended September 30, 2026, up from about $335.8 billion three months earlier. Its TSMC holding was worth $8.64 billion, or 2.48% of the portfolio, and included about 18.93 million shares, nearly unchanged from the prior quarter. TSMC’s October revenue rose 59.7% year over year to about NT$586.9 billion, while revenue for the first ten months of 2026 was up 43.3% to about NT$4.49 trillion. The article notes risks from TSMC’s concentration of its most advanced production in Taiwan, the high cost of building fabs elsewhere, and the possibility of slower AI orders.

AI’s promised productivity boom has yet to benefit workers

Artificial intelligence is changing U.S. workplaces through tools such as automated meeting summaries and help with radiology diagnoses, but the productivity boom promised by technology leaders has not yet appeared. Past technology-driven productivity gains have often taken years or decades to lead to higher pay, and worker compensation has recently grown more slowly than productivity. Federal Reserve economists say AI’s effect on overall productivity growth appears small so far, while some experts see risks in the optimistic predictions. These issues were discussed at the Fed’s Productivity and Jobs Task Force Conference in San Francisco on November 16, where Fed Chair Kevin Warsh said policymakers must plan for a rapidly changing world.

Mark Zandi says AI drives two thirds of US investment growth

Moody’s Analytics chief economist Mark Zandi says artificial intelligence has driven nearly two thirds of US investment growth since ChatGPT launched four years ago. He warned that investment growth in other parts of the economy has slowed by more than half since mid 2024. Zandi said the AI boom supports economic growth and job creation, but also raises electricity and consumer electronics prices, pushes up interest rates, and takes investment away from other areas. He said the investment in AI needs to lead to higher productivity because the rest of the economy is not investing enough to do so.

Spin.AI brings browser extension risk checks to Microsoft Edge for Business

On November 19, 2026, Spin.AI made its browser extension risk assessments available in the Microsoft Edge for Business management service. The assessments draw on analysis of more than 1,000,000 browser extensions and applications, using more than 25 risk factors and checks for vulnerabilities, suspicious code, and unusual behavior. Spin.AI says 50% of extensions in use are high risk, and organizations with more than 2,000 employees use an average of 1,454 extensions. The company also cited the 2025 RedDirection campaign, in which malicious updates to extensions affected more than 16 million users. Spin.AI founder and CEO Dmitry Dontov said the information can help security teams review extensions when deciding whether to allow them.

Simple prompt search matches costlier AI optimizers on key tasks

A study introduces Demonstration-Conditioned Prompt Search, or DCPS, to compare prompt optimizers without giving them extra advantages from larger sampling budgets. DCPS uses randomly selected demonstrations to create prompts, scores them on validation data, and keeps the best without learned updates, reflection, or evolution. It matched StablePrompt’s classification accuracy within 0.5 percentage points at about one third of the measured cost. In DSPy-based AI systems, it matched GEPA’s macro accuracy on GPT-4.1-mini while using 55% fewer rollouts and 32% lower average measured API costs. The researchers found that simpler prompt search performed well on structured and instruction-following tasks, while more complex methods still helped on some open-ended and reasoning-heavy tasks.

Gabon signs three Huawei deals on AI, security and education

Gabon signed three memorandums of understanding with Huawei in Shanghai during President Brice Clotaire Oligui Nguema’s visit. The agreements cover intelligent public security systems, digital modernization at public universities, and AI teaching equipment for Brice Clotaire Oligui Nguema University. The public security agreement provides for the use of Huawei’s intelligent technology products in Gabon. The agreements were signed in the presence of the president and other senior officials. They build on Gabon and Huawei’s technology partnership and aim to support the country’s digital economy and technology capabilities.

Nvidia’s AI financing guarantees rise to $118 billion

Nvidia’s maximum potential exposure from financial guarantees tied to AI infrastructure rose by $9.5 billion from the previous quarter to $118 billion. The largest portion, $105 billion, supports SB Energy’s PORTS-Pike Technology Campus in Ohio, which is expected to host Nvidia systems under long-term leases to OpenAI. The guarantees take effect only after each building is completed and ready for service, with the first expected to qualify in fiscal 2029, and they decline over the 20-year lease terms as OpenAI makes payments. Nvidia CFO Colette Kress said companies must arrange financing and complete construction before guarantees apply, and that Nvidia had not introduced a new financing policy. The company also reported $32 billion in future commitments tied to neocloud agreements, separate from its guarantees, and said it would remain disciplined in its financial commitments.

PACT adapts time series models while preserving past learning

Researchers propose PACT, a method for adapting pretrained time series foundation models to changing operating conditions. It divides the model’s embedding space into local regions and learns small adjustments only where conditions cause changes, while keeping an identity path and earlier adjustments in place. The method uses just 5% labeled calibration data for each new condition and does not need to know the condition during testing. In a four-condition bearing fault test, PACT kept average macro-F1 above 0.95, with similar results on public benchmarks involving matching task and label types.

Investors question whether Big Tech can withstand AI competition

Large technology companies still dominate computing infrastructure, AI services, and consumer applications, and they use strong cash flow and low-cost debt to fund AI investments. But recent earnings from Alphabet and Amazon showed unexpected weakness in some core businesses, raising concerns that AI competition could hurt established companies. Supporters argue that even a serious AI downturn would be manageable because these firms have strong cash flow, high profit margins, and other sources of support, including investments in companies such as Anthropic. If investors become less confident in the companies’ ability to withstand disruption, they may shift money away from hyperscalers despite their financial strength.

Wilgot raises $3 million to help products show up in AI shopping

Wilgot raised a $3 million seed round to develop its platform and expand its commercial reach. The company uses retailers’ search queries, reviews, support tickets, and returns to improve product pages, feeds, FAQs, and structured product data for AI shopping tools. Wilgot says net sales rose 20% across Daniel Wellington product pages it optimized, while 7% of Myrqvist customers now arrive through ChatGPT; the announcement did not provide measurement periods or experimental details. Its customers also include J.Lindeberg and Signify. Anders Nygren of Foundry Ventures and Andreas Söneby of Industrifonden will join Wilgot’s board, and the company says its platform cannot control recommendations made by independent AI assistants.

Fensory adds controls for AI agents trading across markets

Fensory connects AI agents to real market trading through Hyperliquid, according to the article. The platform aims to support crypto, stock-linked markets, gold, oil, and forex through one account. Users can choose which markets an agent can access and set order-size limits, which Fensory checks before submitting an order. The article notes that these controls do not remove the risks of automated trading, leverage, or losses.

Sources

NOTE:

This news brief was generated using AI technology (including, but not limited to, Google Gemini API, Llama, Grok, and Mistral) from aggregated news articles, with minimal to no human editing/review. It is provided for informational purposes only and may contain inaccuracies or biases. This is not financial, investment, or professional advice. If you have any questions or concerns, please verify all information with the linked original articles in the Sources section below.

TSMC AI chip manufacturing ASML Nvidia AI infrastructure AI accelerators AI investment AI productivity Big Tech AI competition enterprise security browser extension risk AI retail AI shopping AI trading agents prompt optimization time-series adaptation Huawei Gabon digital infrastructure AI economics

Comments

Loading...