TSMC's US Plant and White House's Military AI Plan
26 Oct 2024 (26 days ago)
TSMC's Arizona Plant Achieves Superior Yields and Its Implications
- The first US chip plant in Arizona, operated by Taiwan Semiconductor Manufacturing Company (TSMC), has achieved yields that surpassed those of equivalent plants in Taiwan, marking a significant breakthrough for the US expansion project (1m55s).
- The yields refer to the volume or number of usable chips in a batch, which is crucial for profit and efficiency, and TSMC's success indicates that they have overcome the early stages of production quickly (3m3s).
- The US government is making a big bet on TSMC, a foreign company, to drive the country's chip production ambitions, and the success of the Arizona plant is seen as a positive sign for this initiative (2m35s).
- The US focus on bringing production to the country may lead to more TSMC production and investment from the federal government, as the company is seen as a key player in achieving this goal (3m39s).
- The timing and manner of the announcement are interesting, as the US needs a win in the chip production initiative, and TSMC may be seeking to remind the world of its capabilities (4m18s).
- Onshoring is one part of the US strategy, but control is also a crucial aspect, as the country seeks to reduce its dependence on foreign companies for critical technologies (4m48s).
- The Biden policy of implementing tough export controls on critical technologies and providing subsidies to companies like TSMC, which received $6 billion in subsidies and $5 billion in loans, is working to push the US semiconductor project forward (5m33s).
- The US focus on onshoring chip supply is graded as an A- due to the country's efforts to stay ahead of China in key technologies, with a focus on slowing down China's advancements through export controls and speeding up US innovation (6m9s).
- The US has implemented tough export controls on advanced chips and tools, but making these controls even tougher unilaterally may not slow down China and could harm US companies' revenue and research and development (6m41s).
- TSMC has admitted that its chips are being used in Huawei's products, despite Huawei being cut off from US business dealings since 2020, highlighting the difficulties in implementing effective limitations on semiconductor expertise (7m14s).
- The limitations on semiconductor expertise can be difficult to enforce, and relying solely on controls may not be effective, as companies can find ways to circumvent them (7m43s).
- Running faster and innovating is considered a more effective approach than relying solely on controls, with the US having an amazing innovation engine that needs to be kept running and funded (8m42s).
- The Chips Act is not only about funding chip production but also about investing $170 billion in basic science R&D funding, which is crucial for creating future technologies (9m7s).
- China is taking its own action and innovating in areas like artificial intelligence, language models, surveillance, and drones, with some of its companies being excellent and first-rate (9m37s).
US-China Technology Race and Competition
- The US and China are in a technology race, particularly in areas like AI, semiconductors, and 5G, with China aiming to advance in these technologies and have a dominant military, which is worrying to the US, Europe, and their friends and allies (10m51s).
- Despite the competition, it is possible for the US and China to do business together in other areas, such as the consumer sector and agriculture (11m31s).
UK's AI Strategy and Investment
- The UK Prime Minister is positioning himself as a champion for AI, calling it an opportunity rather than something to be scared of, and has appeared alongside the former Google CEO to promote this view (12m27s).
- The UK's AI plan is not yet clear, but it is expected to be a twofold strategy, with one part focusing on using AI for improvements in the public and private sectors, and the second part involving legislation around AI (12m55s).
- The UK government has promised not to go as far as the EU in regulating AI, and has given rough commitments to make voluntary commitments into law and establish a new agency to monitor and enforce safety and criteria for AI labs (13m11s).
- The UK government has had to navigate and educate the public on AI, including nixing a couple of portent investment areas, and is reevaluating the computing budget for the country (14m10s).
- The amount of money going into AI and innovation in the UK is small compared to the US, with concerns that the UK is losing its footing to France, which has invested quite a bit in this area (14m59s).
- The UK government has been working to prop up the industry, but defenders of the government argue that more needs to be done to support AI and innovation in the country (15m23s).
- The U.K. government has been discussing its ambitions to become a science superpower, with former Prime Minister Rishi Sunak mentioning the goal, but the progress has been slow, and the government often uses buzzwords like efficiency without concrete actions (15m48s).
- The U.K. government has been having conversations with the startup industries, but there is limited evidence of them being drawn into government contracts (16m16s).
US Government's Push for AI Adoption in Military and Government
- The White House has announced that it needs AI adoption in the military to maintain an edge over rivals and increase safety (18m3s).
- The primary reason for the U.S. government's push for AI adoption is that its adversaries are already deploying AI, and the U.S. needs to keep up to maintain its values (18m29s).
- The Biden administration has published a memo outlining a strategy for AI adoption in the military and government, but the existing administration has limited time to implement it due to the upcoming presidential election (19m9s).
- The Republican platform has stated that President Trump would repeal many of the Biden administration's consumer policies on AI, but it does not have a clear take on military and intelligence uses (19m29s).
- Kamala Harris has been involved in developing AI policies on the international stage, presenting a different vision for the future of AI (19m47s).
- The strategy for AI adoption in the military and government is based on working with the private sector, with companies like Palantir already working closely with the government (20m11s).
- James Sullivan, President Biden's National Security Advisor, has emphasized the importance of working with the private sector to advance AI adoption in the military and government (20m13s).
- The US government is working with tech companies to advance innovation, with a focus on collaboration between the public and private sectors to drive technological progress (20m27s).
- The government is behind in terms of technology, and there is a need for cooperation between the two to make advancements work (20m53s).
- There are ethics concerns and limitations from the United Nations regarding the use of autonomous weapons and AI in the battlefield (21m2s).
- The Biden administration has developed a risk-based approach to AI in military and intelligence settings, with the riskiest uses requiring the most safeguards and ethical considerations (21m16s).
- The administration is thinking about AI risks in terms of tiers, matching their approach to consumer AI, and is working to expand the database of vendors for this technology (21m41s).
- The government is trying to work with both large companies and startups to engage with the community and provide access to opportunities to help the US government and military (22m15s).
- The tech industry is driving the NASDAQ to a new record high, with companies like NVIDIA and Tesla leading the way (23m24s).
- Apple is down on the week but up on the day, with a key analyst report from KeyBanc drawing attention to potentially overly optimistic views on the company (23m49s).
- Apple has a series of events planned, including the rollout of new Mac models with the M4 chip designed to support AI capabilities, and is expected to create a mega news cycle ahead of the holiday shopping season (24m12s).
- The new Mac models, including the Mac Mini, are due for a drastic design refresh, with the device getting much smaller than it is currently (25m6s).
- Apple's focus on artificial intelligence is expected to be a key driver of the company's progress, with anticipation building ahead of the company's earnings report (25m21s).
- KeyBanc Capital Markets analyst has expressed concerns that the market's expectations for Apple's growth are unrealistic, assuming the company will grow across all product lines and regions simultaneously, which statistically happens only once every few years (25m43s).
- In China, the sales data for the iPhone 16 has been inconsistent, but early numbers have been encouraging despite soft sales of luxury goods in the country (26m45s).
- The upcoming earnings report may not shed much light on iPhone 16 sales specifically, but there will be scrutiny on China sales (27m11s).
Market Outlook and Magnificent Seven Earnings
- Thomas Martin, Senior Portfolio Manager at Global Investments, is optimistic about the earnings reports of the "Magnificent Seven" companies, which have strong fundamentals driving their revenues and earnings (27m40s).
- The Magnificent Seven companies are expected to report earnings growth of around 19%, significantly higher than the S&P 500, but also the slowest rate of EPS growth for six quarters (28m33s).
- Thomas Martin notes that the market has had a big run-up to a new high and is now above that high again, but may need to take a breather after companies have risen strongly (29m41s).
- Thomas Martin still likes semiconductor companies, particularly those that are part of the supply chain for companies like TSMC (30m5s).
- The outlook for NVIDIA and the demand for their products, driven by the need for large language models in AI, is still in its early stages, with fundamentals supporting high stock prices (30m19s).
- NVIDIA's earnings report on November 21 will be closely watched, with signals from companies like Amazon, Apple, and Meta on their spending on NVIDIA's chips (30m47s).
- The CAPEX numbers in NVIDIA's report will be critical, as they will indicate the level of commitment and investment in the company (31m12s).
- It's too early for NVIDIA to back off on capital spending, and the company is unlikely to cut back on costs unless there's a slowdown in revenue and business (31m19s).
Crypto Companies' Investment in US Election and Candidates' Stances
- Crypto companies are making massive investments in the upcoming US election, with the goal of measuring politicians' takes on the industry (32m24s).
- Crypto backers have directed more money to support their industry than any other interest group, with the super PAC Fair Check leading the way (32m45s).
- Coinbase predicts that political financing will translate to regulatory results, and a pro-crypto Congress will emerge regardless of the outcomes in individual races or the presidential race (33m15s).
- Presidential candidate Donald Trump has changed his stance on crypto, now supporting it and promising to make the US the "crypto capital of the planet" (33m40s).
- Vice President Kamala Harris strikes a potentially softer stance on crypto, with her campaign's crypto plan being more vague and focused on protecting participants in the market (34m9s).
- Crypto insiders believe the industry will outlast the election, and that it should not be partisan (34m37s).
- Activate Consulting has released its 2025 Tech and Media Outlook Report, which forecasts the technology, internet, and media sectors to add over $1.1 trillion in global growth over the next four years (36m28s).
- The report's CEO, Michael Wolf, notes that the consumer side of the business is expected to see around $400 billion in incremental revenue, while the B2B side is expected to see an additional $700 billion in revenue, growing three percentage points faster than the consumer side (37m10s).
- The report's forecasts are based on a year's worth of research, including consumer and B2B research, and taking apart the plans of major companies to identify underlying drivers of growth (37m45s).
- Generative AI is expected to play a significant role in driving revenue growth, particularly in the enterprise sector, where companies will see increased revenue from pricing for additional penetration of their B2B applications (38m59s).
- However, there is some skepticism about the return on investment for AI, with the IBM CEO noting that it is still costly and not easy enough for enterprise use (39m43s).
- Michael Wolf acknowledges that there is a hype cycle around AI, but notes that there will be specific applications that will drive growth (39m51s).
- Waymo has raised $5.6 billion in its largest ever funding round, led by Alphabet and including investors such as Andreessen Horowitz, Fidelity, and T. Rowe Price (35m42s).
- China's WeRide has raised over $44 million in a delayed IPO, pricing at the bottom of its arranged range of $15.5-$18.5 (35m58s).
- Agility Robotics has raised $150 million from investors, led by investment firm DCVC, valuing the maker of humanoid robots at around $100 (36m17s).
- The impact of generative AI on companies is unlikely to provide huge amounts of additional revenue, but rather increased functionality and better tools (40m6s).
- The focus is on predicting what will happen in terms of specific sales of existing products rather than inventing what could be coming (40m27s).
- The AR market is a nascent area, with 54 million headset shipments forecasted by 2028, which is considered underwhelming at 4% of smartphone sales (40m49s).
- However, more sophisticated headsets are expected, and AI is already ahead of the device in this space (41m7s).
TSMC's US Plant Success and Market Demand
- TSMC's US plant has achieved superior yields, even surpassing those produced at home, which is a significant development for American chip manufacturing (42m8s).
- The overall capacity of TSMC's 5-nanometer production accounts for 35% of their overall revenue, giving them pricing power due to their leading-edge technology (42m22s).
- This could lead to hyperscalers like NVIDIA, Apple, and Microsoft partnering with TSMC to produce their latest products onshore in the US (42m47s).
- The market expects high demand for TSMC's US-produced chips, with potentially 20,000 wafers sold out on day one (42m57s).