EP71 — Unicorn Files — Skeleton Technologies: what if AI’s bottleneck is no longer chips?

An Estonian-founded deep-tech company spent seventeen years building high-power storage technology before AI data centers emerged as a major target market. This episode examines the real physics of transient electrical loads, Skeleton’s GrapheneGPU offering, the Siemens investment and exclusive global data-center partnership, competing approaches and the gap between performance claims and independently demonstrated commercial value.

• Rapid GPU-cluster power swings are a real problem, distinct from total energy availability.
• Supercapacitors can buffer transient loads but do not create electricity or intrinsically accelerate chips.
• Siemens announced a strategic investment and exclusive global data-center commercial partnership.
• A valuation above $1 billion is not itself proof of profitability or commercial scale.
• Skeleton’s quantitative performance claims require independent benchmarks and deployment data.