Heron Power

Overview

Heron Power is a privately held power-electronics company founded in 2025 by Drew Baglino, a former Tesla senior vice president, and based in Scotts Valley, California. It designs and intends to manufacture solid-state transformers and medium-voltage power electronics for AI data centers, solar and storage projects, and the grid. The company is pre-product, targeting roughly 10 engineering prototypes in summer 2026 and pilot production in early 2027.

The core pitch is about the transformer itself. A conventional power transformer is a passive device of copper windings and a steel core that has changed little since the Edison era. It steps voltage up or down at grid frequency and does nothing else. Heron's argument is that this device should be replaced with a controllable one built from power semiconductors, able to convert and regulate power rather than only change its voltage.

Timing is the other half. AI data-center construction has driven up demand for grid-connection equipment while conventional transformers are in short supply, with multi-year lead times on large units. Heron positions Heron Link, its SST product, as a faster-to-deploy and more capable substitute. None of this is proven in the field. The claims below are the company's, and the product has not shipped.

The technology

A solid-state transformer does the job of a conventional transformer using power electronics in place of a passive electromagnetic core. A conventional transformer steps AC voltage up or down through magnetic coupling between copper windings on a steel core. It is reliable and cheap, and it cannot control or convert power. It changes voltage at line frequency, and converting between AC and DC or regulating voltage and frequency requires separate equipment alongside it.

An SST uses semiconductor switches operating at high frequency. Switching thousands of times faster than grid frequency shrinks the magnetics, which reduces size and weight. The same electronics let it convert between AC and DC, push power in both directions, and regulate voltage and frequency directly, folding several functions into one device. Modern designs use wide-bandgap semiconductors such as silicon carbide, which handle higher voltages and switching speeds with lower losses than older silicon parts.

This is established power-electronics theory, not unique to Heron. SSTs have been researched for years and remain an emerging product category at medium-voltage scale, with limited commercial field deployment across the industry.

Heron's specific claim is the Heron Link unit. The company says it connects directly to 34.5 kV medium voltage and outputs 800 volts DC, built on wide-bandgap semiconductors, at a claimed conversion efficiency above 98.5%. In Heron's design, a single Heron Link replaces both the legacy MV transformer and the separate regulation and conversion equipment that normally sits beside it. The efficiency and capacity figures are design targets, not independently measured results.

Products

The 800 VDC data-center blueprint

In May 2026, Heron published a Blueprint for 800 VDC Datacenters laying out how it would power an AI facility with Heron Link units. The building block is a 12 MW IT block treated as a repeatable unit, powered by four Heron Links in a four-to-make-three redundancy arrangement: three carry the load, the fourth is spare. A single Heron Link plus SuperBBU is meant to replace four equipment categories at once, the MV transformer, the low-voltage switchboard, the uninterruptible power supply, and the power distribution unit. Heron argues this collapses the conventional MV-to-rack chain and removes the dedicated electrical room.

The blueprint carries a set of headline numbers, all of them Heron's own modeled comparison against a traditional 480-volt AC design, with the baseline drawn from Schneider Electric and McKinsey references. In that comparison Heron claims roughly 65% lower MV-to-rack equipment cost, about 90% lower install labor, a 25% faster construction schedule, 50% lower power-delivery losses, 30% less copper, about 2.5 times the power density per square foot, and a single procurement lead time of roughly 24 weeks in place of multiple long-lead items.

These are modeled figures from the company, not independently audited results, and Heron has no shipped product behind them. They describe what the design is meant to achieve and should be read as a vendor's projection rather than measured field performance.

Funding & investors

Heron raised a $38 million Series A in May 2025, led by Capricorn Investment Group through its Technology Impact Fund. Other investors were Breakthrough Energy Ventures, Energy Impact Partners, Gigascale Capital, Powerhouse Ventures, and Valor Equity Partners, along with angel investors JB Straubel, a Tesla co-founder, and Zach Kirkhorn, Tesla's former chief financial officer. Straubel and Kirkhorn are investors in Heron, not operators or executives. Total disclosed funding stood at about $43 million after the round.

In February 2026, Heron closed a $140 million Series B co-led by Andreessen Horowitz through its American Dynamism fund and by Breakthrough Energy Ventures. Capricorn, EIP, Gigascale, and Valor, through its Atreides AI Fund, also took part. That brings disclosed funding across the two rounds to at least roughly $178 million. Heron has not disclosed a valuation.

The backing is heavy on Tesla alumni and climate-focused venture capital, which signals investor conviction in the team and the thesis. It is not, on its own, evidence that the product works at scale.

Manufacturing & traction

Heron plans a 40 gigawatt-per-year U.S. manufacturing facility funded by the Series B. In August 2026 it named the location, Morgan Hill, California, about 30 minutes south of San Jose, where it is converting a 286,000 square foot former distribution warehouse into a plant it calls Heron Factory One. Heron puts the investment at over $100 million and the headcount at more than 600. The plant is designed to build up to 10,000 Heron Links a year at 5 MW each, which is where the 40 GW figure comes from. Site preparation began in August 2026 and mass production is slated for late 2027. TechCrunch estimated that 40 GW per year would be roughly 10% to 15% of annual transformer production outside China.

In May 2026, the California Governor's office announced a $26,375,000 California Competes Tax Credit for Heron, tied to a stated company investment of $140.9 million and 601 new jobs. The credit is a performance-based award against those commitments, not cash paid up front.

The Series B followed about 40 GW of expressed customer interest, which Heron has cited from more than a dozen prospective customers in active technical collaborations. This is interest, not booked or contracted orders. The distinction matters: it reflects demand signals and engineering engagement, not signed purchase agreements or backlog.

On timeline, Heron is targeting roughly 10 engineering prototypes in summer 2026, deploying some with early customers, pilot production in early 2027, full-scale production in the second half of 2027, and a ramp over roughly the following two years.

Partnerships

Heron was named in NVIDIA's Open Innovation Ecosystem for 800 VDC data centers, announced at the Open Compute Project Global Summit in October 2025. The same list includes incumbents such as ABB, Eaton, GE Vernova, Hitachi Energy, Mitsubishi Electric, Schneider Electric, Siemens, and Vertiv. This is ecosystem alignment around a common 800 VDC architecture, not a supply contract or a commitment by NVIDIA to buy Heron's equipment.

In June 2026, Heron and LG Energy Solution Vertech announced a design and engineering collaboration to integrate Heron Link with U.S.-made battery storage. As described, this is a joint engineering effort, not a firm purchase agreement.

Leadership

Heron's founder and chief executive is Drew Baglino. He spent roughly 17 to 18 years at Tesla, most recently as senior vice president of powertrain and energy engineering and one of the company's named corporate officers, and departed in April 2024. The departure date comes from general reporting rather than Heron's primary releases.

No co-founders or other executives are named in public sources. Baglino is the only named leader Heron has disclosed.

Strategy & outlook

Heron's bet is that the AI buildout and the shortage of conventional transformers create a durable market for a controllable, integrated, faster-to-deploy medium-voltage product, and that owning the full stack, the SST, the SuperBBU, and a 40 GW factory, is the way to capture it. The integrated approach is meant to let one device replace several, shorten procurement lead times, and give data-center and renewables developers a single connection product.

The company still has to prove the product in the field and execute a manufacturing ramp at a scale no SST maker has reached. The thesis is coherent and the demand backdrop is real. The open question is execution, from first prototypes through Heron Factory One, the 286,000 square foot Morgan Hill, California plant Heron named in August 2026 and does not expect to reach mass production until late 2027.

Key considerations

Heron is pre-product. Every performance and cost number it has published is a design target or a modeled claim, and nothing has shipped or run in a customer's facility. The efficiency, the per-unit rating, and the 800 VDC blueprint describe what the company expects its hardware to do, not what measured units have done.

The customer pipeline is the second point. The roughly 40 GW Heron cites is expressed interest and technical collaboration, not contracted orders or backlog. Demand interest at this stage does not convert one-to-one into revenue.

The third is manufacturing. A 40 GW-per-year SST factory would be a first of its kind at that scale. Heron has now named the Morgan Hill site and put over $100 million behind it, but it has disclosed no capital plan beyond that figure, the Series B, and the California Competes commitments, and the building has to be converted and ramped before it produces anything. Mass production is not due until late 2027. First-of-a-kind manufacturing ramps in power hardware have a long history of delay and cost overrun.

SSTs are commercially immature at medium-voltage scale. Certification, utility acceptance, long-term field reliability, and standards for connecting these devices to the grid are unsettled across the industry, and Heron has to clear those gates. It also faces well-capitalized incumbents, the same ABB, Hitachi Energy, GE Vernova, Eaton, Siemens, and Schneider Electric named in NVIDIA's ecosystem, plus other SST startups. The market statistics Heron uses to frame the opportunity, such as 24-month transformer lead times and less than 20% of U.S. transformer demand met domestically, are the company's own framing and should be read as such. With no disclosed valuation and a single named executive, the company itself is still an early-stage unknown.

Sources