The semiconductor sector is splitting into two distinct valuation tiers. AI-native silicon commands premium multiples; legacy analog and mixed-signal players trade at discount. The divergence is structural, not cyclical.
Nvidia's Vera Rubin, AMD's EPYC Venice, Astera Labs' Scorpio X-Series, and Broadcom's networking ASICs are racing to hyperscaler customers. Data center buildouts are compressing lead times and pushing incumbents to lock in supply agreements years in advance.
ASIC design costs at leading-edge nodes now run into the hundreds of millions of dollars. That barrier is eliminating new entrants before they reach tape-out. Industry quality standards compound the challenge: failures are measured in parts per million, and every anomaly requires root-cause documentation and analysis.1 Academic prototype methodologies — where 5 to 10 functional chips out of 40 suffice for publication — have no place in production environments.1
The geopolitical overlay is accelerating the split. U.S. rare earth import bans are pressuring supply chains built on decades-old assumptions. China's Zhenwu V900 and J900 AI chips signal a parallel technology stack developing outside U.S. export control reach. The industry's integrated global supply chain is being replaced by two competing ecosystems.
Analog Devices illustrates the divergence inside a single balance sheet. Automotive revenue reached 24% of total revenue in Q2, up 8% sequentially and 2% year over year.2 That segment strength masks concern about memory supply chain choke points that could affect consumer customers.2 Automotive provides a floor; AI exposure determines the ceiling.
Beyond digital AI silicon, GlobalFoundries is targeting a new frontier. The company launched Quantum Technology Solutions to scale U.S. quantum manufacturing, partnering with PsiQuantum on photonics-based quantum computing.3 PsiQuantum CEO Victor Peng cited the partnership as critical for utility-scale quantum delivery, pointing to state-of-the-art photonics results as proof of what a domestic manufacturing partner can deliver.3 Quantum remains pre-revenue, but it marks where the next valuation premium may form.
For equity investors, the framework is clear. Incumbents with AI silicon in production — or firm hyperscaler commitments — trade on forward demand. Companies without direct AI exposure face multiple compression regardless of legacy strength. Supply chain geography now affects addressable market, not just margins.
ASIC cost barriers, geopolitical fragmentation, and AI infrastructure spending cycles all point to further separation between the two tiers.
Sources:
1 Anonymous ASIC Designer, IEEE Spectrum, May 28, 2026
2 Analog Devices Inc, NewsEOD, May 20, 2026
3 Victor Peng / GlobalFoundries, GlobeNewswire, May 21, 2026
