The future of LLMs in the AI race

Key Concepts

• S&P 500: A stock market index representing the performance of 500 of the largest companies listed on stock exchanges in the United States.
• Artificial Intelligence (AI): The simulation of human intelligence processes by machines, especially computer systems.
• NVIDIA: A technology company known for its graphics processing units (GPUs), which are crucial for AI development and computing.
• Taiwan Semiconductor Manufacturing Company (TSMC): The world's largest contract chip manufacturer, producing chips for companies like NVIDIA.
• Large Language Models (LLMs): A type of AI model trained on vast amounts of text data, capable of understanding and generating human-like text.
• Capital Outflow Needs/CAPEX Spending: Expenses incurred by a company to acquire, maintain, or upgrade its physical assets, such as property, buildings, and equipment.
• Consumer Sentiment: A measure of the overall attitude of consumers toward the economy and their personal financial situation.
• Interest Rates: The cost of borrowing money or the return on lending money.

Market Performance and AI Enthusiasm

The S&P 500 is experiencing a rise today, primarily driven by an increase in Alphabet shares. This positive movement follows a four-day decline in the technology sector, indicating a renewed investor interest in the artificial intelligence (AI) market. Investors are anticipating NVIDIA's upcoming earnings report, hoping it will alleviate concerns that AI stocks are currently overvalued and overhyped.

NVIDIA's Earnings and Future Outlook

Kim Forrest, Chief Investment Officer and Founder at Bokeh Capital Partners, projects a strong quarter for NVIDIA. She notes that the performance of Taiwan Semiconductor (TSMC), NVIDIA's chip manufacturer, which reported about a month ago, provides a good indication of NVIDIA's own results. TSMC's ability to earn more from complex chips suggests a positive outlook for NVIDIA.

However, the key focus for investors will be NVIDIA's bookings for the next six months. Forrest believes it is still too early in the AI build-out phase to expect a rapid decline in demand. While acknowledging the "secret fear" among investors of a sudden downturn for NVIDIA, she does not anticipate this happening in the current quarter, leading to a "sanguine" outlook.

Forrest also suggests that the broader AI trade has contributed to recent sell-offs in US markets, as investors express nervousness about the future landscape of AI.

Perspective on Large Language Models (LLMs)

Forrest expresses reservations about Large Language Models (LLMs), stating, "I am not sure that currently we can make that leap to have them be thinking machines." Drawing from her experience as a former software engineer who worked on early neural network software, she argues that thinking involves more than just pattern matching and statistical analysis, which are the core functions of current LLMs. She believes LLMs are excellent at understanding user intent but not yet capable of true thought.

Preferred AI Applications

Instead of a "one model to rule them all" approach, Forrest favors smaller AI models that address specific, short-term problems. She cites Palantir as an example, acknowledging that while the company is "extremely overpriced," it effectively solves smaller, well-defined problems with AI, avoiding the need for a significant conceptual leap. She prefers AI applications that enhance productivity and is uncertain about the tangible problems LLMs are currently solving, making it difficult to determine a fair valuation for their access.

Concerns Regarding Bond Issuances

A significant concern for Forrest relates to bond issuances from companies like OpenAI. She highlights that these companies, despite having substantial revenue, may not have revenue streams that match their capital outflow needs or capital expenditure (CAPEX) spending requirements. This creates anxiety for bondholders who are concerned about repayment. The uncertainty surrounding future revenues and user access to these models makes it challenging to accurately assess the risk of bond defaults. This situation, she believes, is prompting a broader investor conversation about the actual value and problem-solving capabilities of these AI ventures.

Tech as the Dominant Market Focus vs. Consumer Sentiment

While acknowledging that tech is the primary focus for investors, Forrest points out the significant weight of companies like NVIDIA within the S&P 500 (approximately 8-9% of the index). She suggests that if the "boil comes off of AI," consumer sentiment, particularly in the housing sector, could become a more significant concern.

She notes that weakening consumer sentiment, as indicated by recent earnings reports from Home Depot and Lowe's, could be a problem. While lower interest rates could alleviate some of these issues, Forrest identifies a larger problem in hiring. She suggests that companies might be prematurely anticipating the use of AI to reduce their need for human employees, potentially leading to a reluctance to hire. This situation, she believes, could be remedied by companies shifting their focus back to hiring.

Conclusion

The current market is characterized by a strong focus on AI, with NVIDIA's performance being a key indicator. While there is optimism surrounding NVIDIA's earnings, underlying concerns about AI stock valuations and the true capabilities of LLMs persist. Forrest advocates for a more pragmatic approach to AI, focusing on specific problem-solving applications rather than broad, speculative models. Simultaneously, potential shifts in consumer sentiment and hiring trends, influenced by both economic factors and AI adoption, warrant close observation.