Independent researcher with a materials science background. Spent 6 months mapping the GaN supply chain before entering a position. Not a professional investor. This is a physics and geopolitics call, not a DCF model. Take the finance mechanics with appropriate skepticism.
GaN Supercycle: Physics, supply chain sovereignty, and the AI compute forcing function
The OpEx pressure is already showing up in earnings calls. Once electricity is your biggest cost, every efficiency point in power conversion is worth pursuing aggressively.
3 is bigger than 1. The efficiency delta is not incremental. It is structural. Silicon cannot close this gap through fabrication improvements alone.
The restrictions are already in place and tightening. FPLSF sits on non-China high-purity gallium. That is not a speculative position. It is a direct response to a documented supply risk.
Enterprise procurement is slow by design. The physics argument wins eventually but 'eventually' can mean 5–7 years in infrastructure timelines.
NVTS is the right bet on the application layer but the timing is the risk. They are spending ahead of revenue and need the market to move on their timeline.
0 propagations
The thesis benefits from the superior efficiency of GaN semiconductors driven by AI compute demand and geopolitical supply chain pressures. Major data center adoptions and China's gallium export restrictions bolster this outlook. However, slower-than-expected GaN technology adoption and economic downturns pose risks. Trusted news sources corroborate these scenarios, supporting a net positive outlook.
The investment thesis relies heavily on the adoption of GaN technology in data centers due to its efficiency advantages over silicon and the increasing demand for power-efficient solutions driven by AI compute scaling.
The thesis focuses on the superior efficiency of GaN semiconductors in power electronics, driven by AI compute demand and geopolitical supply chain pressures.
Related News
Citigroup raised its S&P 500 year-end target to 8,100, projecting a 9% gain driven by real AI earnings growth. SpaceX's anticipated $1.77 trillion IPO is expected to increase market liquidity in summer 2026. The AI supercycle has reached a stage where deployed hardware generates revenue, encouraging continued investment despite semiconductor volatility.
Beijing has imposed stricter licensing on gallium exports, controlling 98% of primary production. This move tightens supply and boosts prices for high-purity material used in semiconductors, benefiting non-Chinese producers.
Bitdeer Technologies mined 921 Bitcoin in May, a 370% increase year-over-year, while its coin balance dropped significantly. The company is focusing on transitioning to AI infrastructure, raising questions about liquidity and sustainability of the pivot.
SpaceX is securing large AI compute deals ahead of its June 12 IPO, including a $920 million monthly contract with Google for 110,000 Nvidia GPUs and related components starting in September 2026. The company also signed a $1.25 billion monthly deal with Anthropic for GPU capacity at its Colossus 1 data center. These deals, which add $26 billion in annual run rate revenue, aim to justify SpaceX's high valuation, though concerns remain about the timing and the AI division's costs.
American Battery Technology (ABAT) shares surged 35% on Monday after the Department of Energy reinstated a $115 million grant for its lithium refinery project, which was terminated in October 2023. The grant was originally awarded in October 2022 and was reinstated following a rigorous review process that concluded in December 2025. The funding supports the first construction phase of ABAT’s Tonopah Flats Lithium Project in Nevada, which aims to produce 5,000 tonnes of battery-grade lithium hydroxide annually. Despite the rally, ABAT stock is still down nearly 30% from its year-to-date high.
Add comment