RockstarMarkets
All news
Markets · Narrative··Updated 1h ago
Part of: Semiconductor Cycle

Memory Chips Rally to Record Valuations Despite Weak Fundamentals: NVDA, AMD Rerating Risk

Insatiable AI demand for memory has pushed semiconductor valuations to all-time highs even as margins compress and supply abundance grows. NVIDIA, AMD and the chipmaking complex are becoming expensive to own on a relative basis, with Bloomberg noting leading stocks 'defy the market's math' as earnings growth lags multiple expansion.

R
Rocky · RockstarMarkets desk
Synthesised from 8 wires · 54 mentions in the last 24h
Sentiment
+45
Momentum
55
Mentions · 24h
54
Articles · 24h
32
Affected sectors
Related markets

Key facts

  • NVIDIA market cap touches $5.5 trillion amid AI infrastructure boom
  • Bloomberg: Top memory stocks now 'cheaper to own' despite soaring share prices
  • Cerebras IPO raises $5.55 billion, priced at top of $23/share range
  • Meta, Amazon, Microsoft diversifying chip suppliers and pushing back on pricing
  • AI buildout shifting from scarcity-driven to efficiency-driven supply model

What's happening

The semiconductor sector has experienced a historic disconnect between valuation multiples and underlying fundamentals as artificial intelligence infrastructure buildout consumes record quantities of memory chips. NVIDIA's market cap has touched $5.5 trillion, with shares up dramatically since early 2023 on the basis that AI capex will remain insatiable. However, Bloomberg research shows these stocks are becoming cheaper to own only in nominal terms; on a price-to-earnings basis, they trade at levels that leave little margin for disappointment.

The culprit is a surge in supply that is beginning to meet demand. TSMC, Samsung and SK Hynix are all expanding capacity, signaling that the memory shortage narrative is fraying. Simultaneously, major buyers including Meta, Amazon and Microsoft are diversifying suppliers and pushing back on pricing, suggesting peak pricing power for chip makers may have already passed. Goldman Sachs analysts have noted that 'the AI buildout is shifting from scarcity-driven to efficiency-driven,' meaning lower-cost alternatives and software optimization will matter more.

This creates a two-tier market: pure-play memory and foundry stocks (NVDA, AMD, AVGO, ARM) face multiple compression if earnings growth slows; meanwhile, AI-consuming companies (META, MSFT, GOOGL, AMZN) benefit from falling capex-per-unit-of-compute. The tape is already showing rotation out of semiconductors into cloud and consumer tech. VIX remains elevated, and breadth on the Nasdaq is deteriorating even as the Mag Seven remain bid.

Bears point to Cerebras' IPO pricing at the top of the range ($23), raising $5.55 billion, suggesting late-stage euphoria in AI chip infrastructure. If macro conditions soften or capex guidance disappoints at NVDA earnings, a sharp revaluation is likely. Cisco's recent beat, however, showed that networking, not just chips, is a bottleneck, potentially extending the cycle and justifying elevated valuations for breadth plays like AVGO.

What to watch next

  • 01NVIDIA Q2 2026 earnings: May 22, guidance on memory demand
  • 02AMD earnings: late May, capacity utilization and ASP trends
  • 03Cisco and AVGO networking demand: validation of infrastructure breadth narrative
Mention velocity · last 24 hours
Coverage from these sources
Previously on this story

Related coverage

More about $NVDA

Topic hub
Semiconductor Cycle: AI Capex, Memory and the SOX Trade

Live coverage of the AI semiconductor cycle — NVDA, AVGO, AMD, ASML, memory demand, capex run rates and overbought signals.