A century-old transformer is a formidable competitor precisely because it is so boring: simple, efficient, durable, and already embedded throughout the electrical grid. That basic conflict gives this technical history a strong foundation. Rather than treating the solid-state transformer as an obviously superior replacement, the presentation repeatedly asks why decades of smaller size, electronic control, and added functionality have failed to displace conventional iron-and-copper designs.
The historical progression is especially useful. The discussion moves from William McMurray’s 1968 patent through U.S. Navy research, Japanese intelligent-transformer work, ABB’s railway prototypes, smart-grid enthusiasm, and the later startup cycle. Those examples establish that solid-state transformers have not merely been an unrealized laboratory curiosity; engineers have repeatedly found plausible applications, demonstrated meaningful advantages, and then encountered cost, efficiency, reliability, or complexity problems. The acknowledgment that some early devices would not satisfy today’s definition of an SST also shows welcome technical caution.
Rail transport provides one of the clearest demonstrations of why the technology has remained attractive. Low-frequency railway power makes conventional transformers especially large, while underfloor installations put a premium on weight and volume. ABB’s reported PET prototypes therefore sound like a strong use case, particularly as the presentation describes efficiencies reaching roughly 95–96% and substantial reductions in size. Yet the account does not mistake prototype success for commercial validation, noting that greater expense, complexity, and limited operating history ultimately kept those systems from widespread adoption.
The smart-grid section adds an equally important economic lesson. Ambitious ideas such as an “internet of energy” made the SST sound like a natural control hub for renewables, electric vehicles, and distributed generation, but utilities proved reluctant to replace equipment capable of surviving for decades with more delicate power electronics. The former Varentec executive’s retrospective is effective because it condenses the commercial problem into a straightforward question: why replace cheap, proven hardware unless the new system delivers value that cannot be achieved otherwise? The later reference to research suggesting large capital-cost reductions would still be required reinforces that point, although many figures and historical claims are presented rapidly without enough methodological detail to assess them independently.
AI data centers make the final case more persuasive because the proposed application is not simply “a better transformer.” The shift toward extremely high-power racks and proposed 800-volt DC distribution creates a specific architectural problem involving current, copper mass, conversion stages, floor space, and power-delivery complexity. The explanation of why higher voltage can reduce conductor requirements is accessible, and connecting SSTs to the possible removal or consolidation of transformers, rectifiers, and UPS equipment makes the economic argument more concrete than the earlier smart-grid vision.
Still, the case remains prospective rather than demonstrated, and the presentation appropriately admits that. Much of the expected benefit currently comes from simulations, roadmaps, investment activity, and planned deployments rather than published operating results from mature 800-volt AI facilities using SSTs. Companies such as DG Matrix, Heron Power, Ampersand, and Delta Electronics demonstrate substantial commercial interest, but fundraising and prototype announcements do not establish efficiency, reliability, maintenance costs, or lifetime economics. The comparison with hybrid architectures based on conventional transformers is particularly important and remains unresolved.
The delivery manages a dense subject well, using trains, naval systems, failed startups, semiconductor advances, and data-center power architecture to turn what could have been a narrow component discussion into a broader technology-adoption story. Humor and informal asides keep the material approachable, although occasional verbal roughness and compressed explanations make some stretches harder to follow than necessary. Most importantly, the conclusion resists declaring victory: after documenting several previous hype cycles, it argues only that AI infrastructure may finally create a narrow commercial opening where the SST’s unusual capabilities justify its disadvantages.
Pros
- The century-long durability and efficiency advantages of conventional transformers are treated seriously rather than being minimized to make the newer technology look inevitable.
- Historical examples from naval research, railway traction systems, smart-grid projects, and failed startups explain why previous SST commercialization attempts stalled.
- The railway discussion clearly connects operating frequency, transformer size, constrained installation space, and the appeal of higher-frequency power conversion.
- AI data-center demand provides a specific potential use case involving 800-volt DC distribution, copper requirements, space, and consolidation of power-conversion equipment.
- Commercial enthusiasm is kept distinct from proven performance, with the lack of real-world 800-VDC SST operating data explicitly acknowledged.
Cons
- Numerous historical efficiency, cost, market-size, and investment figures arrive quickly without enough supporting detail to judge their underlying assumptions.
- Some technical explanations are compressed enough that viewers unfamiliar with power electronics may struggle with the distinctions among SSTs, hybrid transformers, converters, and galvanic isolation.
- The AI data-center argument depends heavily on simulations, architecture roadmaps, announced deployments, and anticipated system savings rather than demonstrated long-term field results.
- Informal phrasing, tangents, and occasional rushed delivery slightly undermine the precision of an otherwise carefully researched technical narrative.
Decades of failed commercialization make the renewed interest in solid-state transformers more convincing, not less, because the presentation shows exactly why earlier markets were insufficient. AI data centers appear to offer a more compelling combination of power density, high-voltage DC distribution, space constraints, and architectural simplification, but the decisive evidence will be operating efficiency, reliability, and cost outside simulations and prototypes. The result is a strong technology history that ends with justified optimism rather than another premature declaration of a revolution.
