NVDA GTC, M&A Wiz / Goog $32 B Deal, April 2 Tariff Uncertainty; Huawei Belt & Road; ChatGPT

The conversation highlights concerns about high structural tariffs and the Biden administration's diffusion rule impacting chip exports, raising questions about America's competitiveness in AI. There are positive implications for Huawei, potentially leading to a "Huawei belt and road." Brad shares insights from a recent visit to Washington, noting positive developments in investments and AI advancements, while expressing optimism for a "golden age" despite current market uncertainties.

Brad discusses the NASDAQ's decline of about 10% since early February, with some stocks down 20-30%. He identifies fears of recession as a second leg of concern, supported by negative data points indicating declining consumer and business confidence. Reports from airlines suggest a drop in consumer demand, as financial concerns lead people to stay home rather than travel. The Atlanta Fed's GDP tracker shows a significant downturn, with the GDP estimate lowered and unemployment rising, indicating worsening economic conditions.

The administration's approach to restructuring globalism is discussed, referencing JD Vance's speech on misconceptions about U.S. production and the impact of cheap labor on innovation. Treasury Secretary Scott Bessett describes the current period as an "American detox," emphasizing the need for fiscal and monetary balance. Anticipation of reciprocal and sectoral tariffs to be announced on April 2 creates market uncertainty, although consumer spending remains strong, with a 6% year-over-year increase in consumer credit growth.

Debate exists over imposing tariffs on semiconductors, considering the AI race and the need for domestic manufacturing. Some argue that tariffs could incentivize the domestic semiconductor supply chain, while concerns about the feasibility of bringing manufacturing jobs back to the U.S. are raised, particularly in light of competition from countries like China. The discussion also touches on automation's potential to replace jobs, raising skepticism about whether current tariff strategies will effectively restore U.S. employment levels.

The conversation reveals that the decline of the American middle class coincided with poverty reduction in China, complicating the trade debate. While tariffs may protect U.S. industries, they could weaken the economy in the long run. Advocates stress the importance of fair competition in global trade, particularly concerning intellectual property and high-paying jobs, as non-tariff barriers from other countries create an uneven playing field. A dramatic increase in tariff revenues could risk pushing the U.S. economy into recession, while a moderate rise could be absorbed more easily.

Market sentiment is currently influenced by tariff concerns, leading to decreased forecasts from the Fed and a cautious approach from investors. Zannie from The Economist notes that European leaders may start to trust China more than the U.S., which could have significant ramifications. However, there are signs of potential investment growth in the U.S., highlighted by the president's announcement of a $1.5 trillion investment and TSMC's $100 billion commitment, including a new R&D facility.

In the M&A landscape, Google has announced a $32 billion all-cash deal to acquire WIZ, a cloud security company projected to generate around $1 billion in annual recurring revenue. This acquisition, while significant, represents a small growth percentage for Google, whose total cloud revenue is about $45 billion. The M&A market is showing signs of recovery, with corporate development teams actively seeking deals, which could indicate a positive shift under the new administration.

The conversation also touches on the challenges of investing in tech companies, particularly in the search space, with Lacework and Wiz mentioned as competitors. The difficulty of achieving success in the tech industry is acknowledged, alongside the value of repeat entrepreneurs in the enterprise sector. The discussion then shifts to the mergers and acquisitions (M&A) approval process, emphasizing its importance in signaling market activity. FTC Chair Matthew Ferguson's stance on M&A is clarified, countering misinformation and highlighting his commitment to regulatory action against clear violations while allowing businesses to thrive.

Attention turns to a recent event featuring Jensen Huang, where his leadership and Nvidia's significance in national and economic security, particularly in AI, are emphasized. The discussion includes the importance of protecting Silicon Valley innovation from competition, particularly from Huawei, and the need for smart immigration policies to maintain a competitive edge. Nvidia's stock performance is analyzed, noting its low valuation compared to the S&P average and concerns about AI demand and competition.

The podcast highlights a significant increase in compute demand, now estimated to be 100 times greater than previously thought, with revised Total Addressable Market (TAM) estimates for AI data centers projected at $2 trillion over the next four years. This surge is attributed to a shift towards machine learning-driven coding, with industry leaders acknowledging the displacement of traditional methods. Nvidia's diverse technology stack and its transition from chip manufacturing to supercomputers are discussed, alongside the impressive capabilities of its new Blackwell architecture.

Despite high demand projections, market sentiment remains cautious regarding Nvidia's growth potential. The conversation concludes with a debate on the scaling of pre-training versus inference, with Nvidia reporting significant efficiency in inference scaling. Analysts observe a market shift, though its significance is debated. Concerns arise over the promotional emphasis on CPU performance metrics, particularly with Cerebras' next-generation chip, Rubin, which is larger and aimed at inference.

The competitive landscape is tightening, with Nvidia's robust supply chain and execution putting pressure on AMD and Intel. AWS is cutting costs on Tranium chips, highlighting competitive challenges, yet Nvidia remains a preferred choice for many, including AWS. The speaker expresses confidence in Nvidia's future, particularly regarding innovations in photonics technology. Tariff impacts on business are discussed, with an upcoming announcement on sectoral tariffs anticipated for April 2nd. The Biden AI diffusion rule complicates chip sales to various countries, raising questions about its implications for Nvidia and the potential risks posed by government actions.

Concerns about Nvidia's market share arise as Wall Street predicts revenue could peak at $250 billion. The discussion touches on the challenges of preventing international startups from leveraging AI models, with criticism directed at an anti-DeepSeek paper from OpenAI, likening it to competitive behavior from traditional industries. High tariffs and the diffusion rule may hinder the U.S. in the AI race, potentially benefiting competitors like Huawei, while U.S. restrictions have prompted China to develop a self-sufficient chip supply chain.

There is a call for a strategic objective to re-establish semiconductor fabs in the U.S., with references to workforce challenges. The conversation shifts to consumer AI demand, with recent launches like Grok and DeepSeek noted. OpenAI currently boasts over 400 million weekly average users and is constructing two large data centers in Texas to support its product launches, although it faces supply constraints in compute power. Bill acknowledges OpenAI's strong position in the competitive landscape, highlighting concerns from rivals like Google, Meta, and Amazon regarding OpenAI's lead in consumer AI.

The discussion shifts to DeepSeek and open-source models, with DeepSeek being more accessible than Llama, which could enhance Meta's competitiveness if made more open. Bill analyzes contribution margins in AI businesses, pointing out that Anthropic has lower margins due to high variable costs, while OpenAI benefits from a direct business model that avoids traffic acquisition costs. He introduces the "girly negative gross margin AI theory," comparing the current AI excitement to the internet revolution, and discusses the complexities of unit economics in AI, including variable costs and high burn rates from VC funding.

Bill emphasizes a "winner take all" mindset in VC markets, which influences pricing strategies, and provides a theoretical example of how negative gross margins could lead to consumers paying less for compute due to subsidies. He predicts potential market resets as companies reassess their unit economics, though the timing remains uncertain. The conversation highlights the importance of understanding top-line revenue numbers and associated costs, emphasizing the need for sustainable business models supported by high-margin revenue. The market is experiencing consolidation and uncertainty, with upcoming weeks expected to reveal more about product developments, tariffs, and tax legislation.