A boom that has almost nothing to do with AI
Every other briefing in this cohort’s future-and-novel-ideas track engages, in some way, with the AI-accelerator buildout. This one deliberately does not. Gallium nitride (GaN) and silicon carbide (SiC) — collectively, wide-bandgap semiconductors — are growing fast for reasons almost entirely independent of AI datacenter demand: electric vehicles, grid electrification, and renewable power infrastructure. It is worth documenting a real semiconductor growth story that is not, in any meaningful sense, an AI story, if only to keep this cohort’s overall framing honest about what is and is not driven by the AI boom specifically.
Why wide-bandgap materials matter physically
Silicon carbide and gallium nitride can handle higher voltages, higher temperatures, and switch faster than conventional silicon power devices, making them structurally better suited to the high- voltage, high-efficiency demands of EV traction inverters, onboard chargers, and DC fast-charging than legacy silicon power electronics [4]. Automotive-qualified 1200V and 1700V SiC devices are now standard components in EV traction inverters specifically [4] — a materials-level shift as consequential to the EV industry as any single battery-chemistry improvement, and one that receives a fraction of the press attention.
Who actually makes these chips
Wolfspeed remains a pioneering pure-play in wide-bandgap semiconductor manufacturing, investing heavily in SiC production capacity specifically to meet growing EV demand [3]. Infineon, covered elsewhere in this cohort’s automotive-and-analog track for its broader power- electronics business, strengthened its GaN position through its 2023 acquisition of GaN Systems, while STMicroelectronics leads alongside Infineon and Wolfspeed in the SiC segment specifically for automotive-grade components [2]. Innoscience and Navitas Semiconductor lead a separate, smaller-device GaN push aimed at consumer electronics rather than automotive traction applications [2].
The bottleneck this boom actually runs into
This cohort’s equipment and materials track documents wafer and materials supply as a recurring chokepoint across the entire semiconductor industry, and wide-bandgap power devices are no exception: 6-inch SiC boule supply from Wolfspeed and Coherent remains insufficient for projected EV ramp volumes, with substrate and wafer suppliers forming the critical bottleneck in wide-bandgap device scaling — leading SiC substrate makers reportedly command premium margins and multi-year customer commitments running through 2028 [3]. Unlike the leading-edge logic bottlenecks this cohort’s foundry track documents at the lithography and packaging stage, this bottleneck sits further upstream, at the crystal-growth stage — SiC boules must be grown slowly and precisely before they can be sliced into usable wafers, a physical process that does not shorten meaningfully with additional capital the way some other capacity constraints can.
Why capital alone can’t shortcut a crystal-growth bottleneck
This distinction matters for anyone reading wide-bandgap growth forecasts and assuming a straightforward relationship between rising demand and rising supply. In lithography and packaging, covered throughout this cohort’s equipment track, additional capital can buy additional tool capacity on a comparatively predictable multi-year timeline. SiC boule growth is closer to a materials-science rate limit: pulling a usable single-crystal boule from a melt takes a fixed amount of time regardless of how many additional furnaces a supplier builds, meaning the 2028 multi-year customer commitments cited above are not simply a matter of insufficient investment, but a reflection of a supply curve that responds to capital more slowly than most other segments of the semiconductor industry this cohort covers.
Why this boom deserves its own name, not a place inside the AI story
Market coverage sometimes folds every semiconductor growth story into the broader AI-boom narrative by association, simply because both are “chips” and both are growing quickly in the same calendar years. This briefing pushes back on that framing specifically for wide-bandgap devices: the demand driving GaN and SiC growth — EV adoption curves, grid modernization spending, renewable energy buildout — would exist, and would be growing, in a world where the AI accelerator boom this cohort covers extensively had never happened. Treating every fast-growing chip category as a downstream effect of AI demand obscures genuinely separate industrial trends like this one, each with its own drivers, its own bottlenecks, and its own multi-decade growth logic worth understanding on its own terms.
