The AI Semiconductor Boom and What Could End It with Stacy Rasgon | The Real Eisman Playbook Ep 63

| Stock Investing | June 08, 2026 | 32.8 Thousand views | 59:06

TL;DR

Bernstein analyst Stacy Rasgon explains how AI infrastructure buildout has made semiconductors the "center of the universe," with the sector rallying 60%+ year-to-date driven entirely by earnings expansion rather than multiple expansion, while Nvidia remains the linchpin whose growth rate dictates the fate of the entire supply chain despite recent underperformance relative to "bottleneck" plays.

🚀 The AI-Driven Semiconductor Rally 3 insights

Earnings expansion drives 60%+ sector gains without multiple inflation

The Philadelphia Semiconductor Index has risen over 60% year-to-date, yet valuation multiples have actually compressed slightly as the rally stems from massive upward earnings revisions rather than speculative multiple expansion.

Memory sector shows extreme earnings leverage

Memory companies like Micron have surged over 100% year-to-date, with SanDisk guiding quarterly EPS for next quarter that exceeds the company's stock price at its IPO just 18 months ago.

AI demand creating sequential bottlenecks across the supply chain

As AI infrastructure scales, constraints have shifted sequentially from accelerators to memory, semicap equipment, optical components, power semis, and CPUs, with investors rotating into each bottleneck area as it becomes the limiting factor.

🎮 Nvidia's Strategic Position and New Disclosures 3 insights

Revenue disclosure reveals 50/50 hyperscale diversification

Nvidia split its data center segment to show revenue is evenly divided between hyperscalers (Google, Meta) and non-hyperscalers (enterprise, neoclouds), addressing concentration risk concerns while consolidating legacy gaming and automotive segments into "edge computing."

Supply chain securing ensures 85% growth acceleration

Unlike constraint plays facing shortages, Nvidia has preemptively locked down supply from TSMC, memory vendors, and advanced packaging partners to deliver accelerating revenue growth from 65% to 85% sequentially.

$20 billion CPU revenue threat to Intel and AMD

Nvidia is targeting $20 billion in CPU sales this year, roughly equal to Intel and AMD's entire CPU businesses, leveraging parallel processing advantages over traditional serial CPU architectures.

⚖️ The 'Constraint Trade' Divergence 3 insights

Nvidia lagging despite being the ecosystem linchpin

While Nvidia's growth determines the health of the entire semiconductor supply chain, its stock has risen only 14% this year as investors favor supply-constrained names like memory and equipment companies.

Market cap creates institutional ownership ceiling

At over $5 trillion and 8% of the S&P 500, Nvidia has become too large for many institutional investors to own beyond market weight, creating technical headwinds despite accelerating fundamentals.

Valuation divergence must normalize

The divergence between Nvidia and its suppliers is unsustainable because the constraint stocks cannot thrive if Nvidia slows, suggesting either the bottlenecks correct downward or Nvidia must re-rate upward to reflect its central role.

🖥️ CPU vs. GPU Architecture 1 insight

Parallel processing dominates AI workloads

While CPUs process computations serially with fewer powerful cores, GPUs utilize thousands of smaller cores for parallel matrix calculations, making them uniquely suited for AI mathematics originally developed for graphics rendering.

Bottom Line

The AI semiconductor trade hinges entirely on Nvidia's continued hyper-growth, creating a temporary market inefficiency where investors can either play the bottleneck constraints or buy the undervalued central enabler before valuations normalize.

More from Steve Eisman

View all