Building a ‘digital twin’ 10,000 feet underground: PNNL, Nvidia and Fervo team up on geothermal AI - GeekWire

In a June 22, 2026 report, the Pacific Northwest National Laboratory (PNNL), chipmaker NVIDIA, and geothermal company Fervo Energy have partnered to develop a 'digital twin' system for geothermal ener... Read original →

This PNNL-NVIDIA-Fervo alliance signals AI’s migration from consumer chips to deep-energy infrastructure. Technically, high-fidelity digital twins demand 3nm-class AI accelerators and EUV capacity, indirectly boosting foundry orders; real-time subsurface simulation also elevates edge compute and low-latency connectivity needs, benefiting smart grid and Sub-6GHz RF chipmakers. On compliance, DOE funding may trigger CFIUS scrutiny if training data includes geological info from regions like Taiwan, China or Southeast Asia, raising Fervo’s overseas operational costs. Competitively, rivals like Quaise Energy could fast-track partnerships with AMD or Intel to develop alternative AI accelerators against NVIDIA’s Omniverse lock-in. Looking ahead, geothermal digital twins will likely become the flagship industrial AI use case by 2027–2028, forcing semiconductor firms to shift efficiency metrics from TOPS/W to Joules per inference—precisely where TSMC’s 3nm HPC platform is heading.