The AI Selloff Doesn't Match the Data | Top AI Investor Explains
Спикеры
GavinSBaker
— Портфельный менеджер, Atreides Management
Gavin Baker dissects the AI and semiconductor drawdown of July 2026, arguing that public market panic sharply contradicts accelerating on-the-ground fundamentals. He explains why GPU spot prices are still surging, how open source models actually boost infrastructure demand, and why hyperscaler operating cash flow acceleration will fund the buildout without a debt crisis. The episode also covers NVIDIA’s new revenue model, the game theory of memory LTAs, SpaceX’s underappreciated compute business, and the looming threat of AI regulation.
- AI names fell 40–60% in July 2026, but every quantitative demand metric (GPU prices, DRAM spot, token growth) is accelerating.
- Open source model adoption takes margin from frontier labs but shifts token share to cheaper tokens, which still consume the same compute—bullish for AI infrastructure.
- Hyperscale operating cash flow accelerated from 28% to 35% last quarter, with contracted compute set to reprice much higher, making the buildout self-financing.
- Memory is the single most important lever for AI throughput; LTAs lock in hyperscalers and give memory suppliers durable pricing power.
- NVIDIA’s credit wrapper/revenue share model effectively creates a cloud royalty stream, reinforcing its competitive position while the stock trades at a 10-year low P/E.
- SpaceX’s public market debut is misunderstood: its compute capacity and ability to bring on power are far ahead of consensus, with significant upside.
- China’s DUV progress is a long-term risk for semicap equipment but an extreme near-term overreaction likely created a watching setup for ASML.
- Regulatory pushback—data center moratoriums and misinformation about water/power drag—is viewed as the biggest potential risk to the AI capex cycle.
Идеи
Hyperscaler cash flow acceleration funds AI buildout
Hyperscale cloud providers (Microsoft, Meta, Amazon) saw operating cash flow accelerate from 28% to 35% in the most recent quarter, even before AI compute contracts reprice significantly higher. As installed GPU capacity reprices at spot rates, operating cash flows will grow substantially, easily funding the AI capex cycle internally and reducing reliance on credit. The selloff on fears of excess capacity and debt-financed overbuild is therefore misplaced; these stocks are undervalued given the earnings power set to materialize.
NVIDIA undervalued on accelerating AI fundamentals
NVIDIA trades at its lowest forward P/E in 10 years, while underlying AI demand metrics (GPU spot prices, DRAM prices, token growth) are accelerating, not decelerating. The market interprets the selloff as evidence that NVIDIA is significantly over-earning, but actual fundamentals show the opposite. NVIDIA’s strategic dominance is reinforced by a new business model (credit wrapper with revenue share) that effectively gives them a cloud-like royalty stream, strengthens competitive moat, and solves the cash flow mismatch for buyers. The stock is deeply undervalued relative to upside from AI infrastructure buildout.
Memory LTA lock-in secures durable pricing
Memory (HBM/DRAM) is the single most important lever to increase token output per unit of compute, making it the axis around which all AI infrastructure revolves. Long-Term Agreements (LTAs) lock hyperscalers into supply; breaking them would destroy the customer’s franchise by risking future allocations. This game theory means memory suppliers have durable pricing power and demand visibility, even in a potential oversupply. The shift from short-term earnings to supply chain agreements is underappreciated, and memory companies are set to benefit from sustained high demand and scarce supply.
ASML selloff on China DUV likely overreaction
China’s reported DUV lithography breakthrough caused a sharp selloff in semiconductor capital equipment stocks. While a domestic DUV capability is a meaningful long-term risk to ASML’s orders, it represents a 2001-era technology, not a near-term threat to EUV dominance. The market likely overreacted, creating a watching opportunity, as any real impact on ASML’s business would be years away and could be forgotten by then. The setup may resolve as a buying opportunity if the initial fear proves exaggerated.
SpaceX compute potential severely underpriced
SpaceX’s AI compute business is massively underappreciated by public markets. The company has already shown it can bring on huge GPU clusters faster and cheaper than anyone, and recent product launches (Grok 4.5, Cursor acquisition) suggest accelerating fundamentals. Consensus revenue estimates of $73 billion for next year look far too low given that a single gigawatt of compute can generate around $50 billion in revenue; reports of 8 GW of capacity would imply triple-digit billions. The market is pricing in a commoditizing spot market and a bearish short case, but execution history suggests the opposite—very little upside to the compute story is in the stock.
Далее
This ILTB Podcast video, published August 04, 2026, features GavinSBaker discussing MSFT, AMZN, META, NVDA, MU, ASML, SPCX. 5 trade ideas extracted by AI with direction and confidence scoring.
Speakers: GavinSBaker · Tickers: MSFT, AMZN, META, NVDA, MU, ASML, SPCX