Nvidia surged roughly 20% after its latest earnings release, and the launch of the Vera Rubin GPU architecture confirmed that hyperscaler capital expenditure on AI compute is not cyclical — it is structural. The move pushed semiconductor ETFs broadly higher and refocused trader attention on the entire supply chain.
Below Nvidia's headline move, a second layer of the semiconductor trade is developing. Edge inference, optical interconnect, and power device companies are positioning for AI workloads that move beyond centralized data centers into automotive, defense, and industrial applications.
On the automotive side, Li Auto adopted Arteris interconnect technology for its L9 Livis SUV. The vehicle's autonomous driving SoC delivers 2,560 TOPS of computing power for real-time AI inference at the edge.1 That figure illustrates the compute density now required in production vehicles — and the supply chain depth needed to support it.
In optical interconnect, POET Technologies reported Q1 2026 results and highlighted its Optical Interposer platform as lower cost, lower power, smaller, and more scalable than comparable solutions.5 A Lumilens purchase order signed earlier this year could scale to $500 million or more in cumulative purchases over five years, per POET's own guidance.5 Lumentum, a larger optical components player, also remains active in data center and AI interconnect deployments.2
Power semiconductors are another active pocket of the trade. Ideal Power is on track to complete deliverables under its existing Stellantis purchase order by mid-2026, advancing its bidirectional power switch technology for EV and industrial applications.4 Blaize Holdings continues to develop low-power edge AI chips targeting inference workloads in embedded and industrial settings.3
Geopolitics is adding urgency to domestic supply chain buildout. A 2027 U.S. ban on Chinese-origin rare earth materials in defense systems forces procurement teams to qualify alternative suppliers for magnets, substrates, and power devices — compressing timelines that previously stretched across years.
For traders, the Nvidia earnings move was the headline. The more durable setup may be the accelerating capital flow into the second tier: edge compute, photonics, and power semiconductors that underpin where AI goes after the data center.
Sources:
1 Arteris, Inc. — May 19, 2026, globenewswire.com
2 Lumentum Holdings — May 18, 2026, nasdaq.com
3 Blaize Holdings, Inc. — May 14, 2026, finance.yahoo.com
4 Ideal Power Inc. — May 14, 2026, finance.yahoo.com
5 POET Technologies Inc. — May 14, 2026, globenewswire.com
