Investors should remain long on the AI theme through the end of 2024, as the current "bubble" is supported by massive earnings growth rather than pure speculation. NVIDIA (NVDA) remains a high-conviction play with a reasonable forward P/E of 25x relative to its projected 80-95% revenue growth. For a value-oriented "catch-up" trade, Intel (INTC) is positioned to benefit from the shift toward Agentic AI, trading at a 30-40% discount to peers like AMD and TSMC. While Micron (MU) will see near-term strength from high-bandwidth memory demand, investors should prepare for a potential 30% to 50% market drawdown in 2025 as semiconductor cycles peak. Avoid traditional software and IT service firms that cannot demonstrate AI-driven productivity, as these sectors face significant disruption and "value destruction" from AI agents.

AI Investment Theme

• The market is currently in a 100% AI bubble, but bubbles form around generational investment opportunities (similar to the internet in the 90s, railroads, or electricity). • Agentic AI (AI that can perform complex tasks across multiple platforms) is a "step function" change in compute demand that started in early 2024. • Corporations are being divided into two categories: those helped by AI and those that are "roadkill." • Token generation (the measure of AI output) is accelerating rapidly, doubling between January and March 2024.

Takeaways

• Don't short the bubble too early: Bubbles can last years. Calling the "pop" prematurely can lead to significant losses while the market continues to climb. • Focus on Productivity: The biggest winners will be companies and individuals who use AI to "supercharge" their productivity (e.g., one engineer managing 10 AI agents). • Watch for "Roadkill": Be wary of companies claiming to use AI while their financial numbers continue to decline; they are likely being disrupted.

NVIDIA (NVDA)

• Despite being the "poster child" of the bubble, its valuation is considered reasonable by some metrics. • Growth vs. Valuation: It is expected to grow revenues by 80-95%, yet trades at a Forward P/E of roughly 25x, which is lower than slower-growing companies like Apple. • Comparison: In the 2000 bubble, Cisco grew at 60% but traded at 140x P/E. NVIDIA’s current fundamentals are stronger than 1990s tech leaders.

Takeaways

• Earnings Underpinning: Unlike the dot-com bubble, current AI leaders have massive earnings and cash flow supporting their stock prices. • Architecture Shifts: The new Vera Rubin architecture uses significantly more memory, which will continue to drive demand for the hardware ecosystem.

Intel (INTC)

• Intel is positioned to benefit from the shift to Agentic AI. • GPU to CPU Ratio: As AI agents become more complex, the ratio of GPUs to CPUs in data centers may shift from 8:1 toward 1:1, benefiting Intel’s core processor business. • Valuation: On an Enterprise Value to Sales basis, Intel trades 30-40% lower than competitors like AMD or TSMC.

Takeaways

• Relative Value Play: Intel is viewed as a "catch-up" trade that is currently under-earning compared to its historical margins and its peers.

Micron (MU)

• High Bandwidth Memory (HBM) is essential for AI, but the semiconductor industry remains highly cyclical. • Historical Warning: In 1995, CEOs claimed the industry was no longer cyclical; the market peaked one month later and dropped 55%. • Technology Shift: While HBM is the current leader, NAND (Solid State Memory) may become more interesting long-term because it "remembers" data when power is off, unlike DRAM.

Takeaways

• Cyclical Risk: Investors should be skeptical of the "it's different this time" narrative. High margins in semiconductors eventually invite oversupply and price crashes. • Near-term Strength: Demand for HBM remains robust for at least the next year due to new AI hardware architectures.

Software & IT Services

• This sector faces significant "value destruction" risks. • Disruption: High-priced consultants and entry-level engineers may be replaced by AI agents (e.g., Claude or ChatGPT). • Budget Constraints: Companies like Uber are blowing through AI budgets (token costs) faster than expected, which may lead to cuts in traditional software spending.

Takeaways

• Bearish Sentiment: Be cautious with traditional software and IT service firms that cannot prove AI is making them more efficient rather than just increasing their costs.

Macro & Geopolitics

• Interest Rates: The market is eyeing Kevin Warsh as a potential Fed Chair who may favor rate cuts by focusing on "trimmed mean PCE" and the deflationary nature of AI. • Oil Prices: High oil prices are a risk, but the administration is incentivized to bring them down ahead of the November elections. • Market Outlook: A 30% to 50% drawdown is predicted for some point in 2025, but the market is expected to "inflate" for the remainder of 2024.

Takeaways

• Election Year Dynamics: Expect the government to attempt to suppress inflation/oil prices to help incumbents, which could provide a tailwind for stocks through year-end. • Prepare for Volatility: While bullish for the short term, investors should have a plan for a significant correction next year.