Engineering brief

The 100x smaller transformer that could rewrite data center power

Weights & Biases1 min read · saves 93 min

At a glance

Relevance
Practical value
Warnings
None

Drew Baglino's Heron Power shrinks transformers 100x by switching at hundreds of kHz. That halves grid-to-chip loss—35 MW more compute per gigawatt.

Power infrastructure is the hidden bottleneck for AI data center buildout.

Summary

Drew Baglino, former Tesla powertrain chief, is building solid-state transformers using wide-bandgap semiconductors. The Heron Link makes the transformer 100x smaller per unit power by switching at hundreds of kHz instead of 60 Hz, drastically reducing the size of magnetic components. This cuts

grid-to-chip losses by roughly half, delivering 35 MW more useful compute from a 1 GW data center. The technical shift is real: wide-bandgap materials (SiC, GaN) allow 10,000V transistors in a die smaller than a GPU. But the operational impact is underappreciated.

Data center electrical infrastructure—transformers, switchgear—is a long-lead, physically massive bottleneck. Baglino's approach replaces multiple cascaded passive stages with a single active power-electronics stage, simplifying buildout. Tradeoffs are significant. Reliability at scale for industrial solid-state transformers is unproven. The grid supply chain is

fragmented and risk-averse. Baglino's claim that data centers will lower electricity rates is plausible but depends on utility rate structures and infrastructure cost allocation. The core insight is that power delivery, not just compute, is becoming a critical constraint for AI scale.

Watch the video

This video is blocked due to your privacy settings. To watch this video, please accept YouTube marketing cookies.

Related breakdowns

Get TL;DW

Too Long; Didn't Watch.

A concise breakdowns of the AI and devtools videos that actually matter for engineering leaders.

Free. Weekly. No hype.

Video and thumbnails remain the property of their respective creators. tldw.news provides editorial analysis, commentary, and discovery links to original content.