Catalyst N° 082 of 125 16 Oct 2025

Calibrating hype with Akshat Rathi

with Akshat Rathi, senior reporter for climate, Bloomberg News, and host of the Zero podcast

In this note

01The question
02The answer
03The argument
04What you need to know first
05Details worth keeping
06Claims worth citing
07Where it’s contested

The question

Which of the energy transition’s hot topics are overhyped, which are underhyped, and how would you tell the difference?

The answer

There is no single verdict. The episode is a scorecard run as a game, and the two speakers disagree about as often as they agree. What emerges from the disagreements is a pattern: the visible objects with quotable lead times, gas turbines and transformers, absorb more worry than they deserve as rate limiters, while delivery of power (transmission) and customer-side capacity (distributed energy resources) get too little. Two rules do most of the work behind those calls. A thing can be important, expensive and genuinely bottlenecked without being the constraint that sets the pace; and hype crosses borders much faster than the talent, supply chains and market rules that would justify it.

03The argument

The trick of the format is that the question is never “is this important.” It is “are we worried about it in the right proportion, and is it the thing that actually sets the pace.” Kann’s answer on transformers is the cleanest example. He calls transformers incredibly important, says the bottleneck is real, says the lead times are very long and that he has seen the problem firsthand from both the utility side and the load side, and then votes overhyped anyway, because transformer lead times are still shorter than gas turbine lead times and projects get built regardless. He runs the same move on generating capacity: long turbine queues are real and widely understood, but there is enough capacity in enough places for the next few years, so it will not be what stops load growth at least through 2030. The verdict tracks how binding a constraint is, not how big or how expensive it is.

The second rule is that a hype rating is only meaningful once you say whose hype you mean, which is why the same topic gets opposite answers from two people who do not disagree about any fact. Kann rates distributed energy resources underhyped while conceding they are becoming hyped inside the wonky energy-tech circles he travels in, where a wave of partnerships between data center operators and DER aggregators is visibly coming. That world is small, and in the broader data center conversation the default is still grid capacity first and a gas turbine second. Rathi reaches the same verdict from the opposite direction: DERs used to be more hyped, as the promised answer to renewable variability, then went quiet, and the products have now actually shipped. Transformers invert that structure exactly. Kann answers from inside the load-growth conversation and says overhyped; Rathi answers for what he calls the non-data-center crowd and says underhyped, because housing projects sit idle for 18 months waiting on one object and whoever can pay jumps the queue. Neither is wrong, and the pair of answers is the closest thing the episode offers to a method.

Rathi’s global lens then overturns several US-formed verdicts, always by the same mechanism: the enthusiasm travels and the enabling conditions do not. Advanced geothermal looks attractive in the US partly because the shale drilling talent and subsurface expertise sit next door, which is exactly why Rathi calls it overhyped everywhere else, since that expertise takes a long time to build. Advanced nuclear runs the same way. The UK is asking Rolls-Royce for a small modular reactor on the strength of American enthusiasm, while the only countries operating civilian small modular reactors are Russia and China, which are also the countries actually building nuclear at home and abroad. The venture capital argument has the same shape from the other side: Rathi calls the VC model overhyped as a way to scale climate solutions, since most of the world’s climate champions were never venture backed, and Kann’s reply is that scaling was never VC’s job. Its job is the first crossing from lab to market, where a very small pool of dollars does outsized work. That defense rests on a counterfactual neither can settle, namely whether BYD becomes BYD without Tesla having existed first, and Kann concedes a narrower point immediately: the investors who fund that crossing often fail to capture its value.

They land hardest on the same side only once, on transmission deliverability, and Rathi’s evidence is deliberately not American. Some 12,000 businesses in the Netherlands are waiting for an electricity connection, and ASML, the country’s most valuable company, cannot build an additional factory for want of one. Kann’s follow-up is the part worth carrying: the Netherlands is not a data center hub, so this queue is not the AI story, which makes it more alarming rather than less. Rathi’s diagnosis for Europe is milder than the American one, since interconnection between countries keeps it from being as severe, but the shape is familiar, with generation in one place, demand centers far away, and transmission failing to keep up.

04What you need to know first

Distributed energy resources (DERs): Kann’s working definition: anything deployed at a customer’s site, on the customer side of the meter, that generates energy, stores it, or shifts load. Rooftop solar, home and building batteries, smart thermostats, EV chargers.
Rate limiter: The one constraint that sets the pace of everything else. Most of the episode’s verdicts turn on separating this from importance: several topics are real, costly bottlenecks that are nonetheless not the rate limiter.
Deliverability: Having generation somewhere is not the same as being able to move it to where the demand is. This is why both speakers can rate generating capacity as overhyped and transmission as underhyped in the same conversation without contradicting themselves.
Small modular reactor (SMR): A reactor small enough to be built repeatedly in a factory rather than assembled one-off on site, the premise being that repetition drives the cost down. Kann’s example is the BWRX-300, a 300-megawatt unit from GE Hitachi.

05Details worth keeping

Rathi’s example of DERs shipping in practice: Octopus Energy, which he describes as the largest utility in the UK, leases BYD electric cars on a monthly plan that includes 12,000 miles a year of free range, on condition that the customer plugs the car in at home so it can be used as part of a virtual power plant. He also gets texts offering free electricity in set windows, such as noon to 2pm.
Kann’s cost-and-uptime argument against co-location: if a data center wants the uptime that cloud facilities have historically had while front-running its grid connection, gas turbines alone will not do it. It needs oversized batteries and extra layers of uninterruptible power supply, at real capital cost. For now the embedded energy cost in compute is small next to the capital cost of everything else, mostly the chips, so cost may not yet bite. He expects the balance between speed to power and cost of power to shift eventually and says he does not know when.
Named co-location examples: xAI’s Colossus data center with a fleet of gas turbines, and a large Meta facility being built with a lot of gas. Kann’s prediction is that co-located generation ends up the minority of new data centers rather than the majority, while still expecting a great deal of new gas capacity overall.
Transformer lead times have been long since roughly 2021 or 2022 and have persisted longer than many expected. Rathi attributes this partly to manufacturer conservatism; Kann agrees and notes the industry has been burned by overbuilding before, so he expects upstart manufacturers rather than only incumbents such as Hitachi to fill much of the gap.
A123 is the awkward case inside the venture capital argument. Lithium iron phosphate chemistry was invented in America and first commercialized by A123, a venture-backed company that went public and then went bankrupt fairly quickly, though some early investors made money. Kann uses it to separate whether venture capital creates value from whether it captures it.
On the Paris Agreement, Kann says he essentially never thinks about it. Rathi’s account is that its effect on business is large but indirect: it became government policy, and government policy shapes what businesses do, and companies attribute their choices to the policy rather than to Paris. COP 30 was due in Brazil the following month.
Kann attributes part of the US hype around advanced geothermal and advanced nuclear to the Secretary of Energy, Chris Wright, being enthusiastic about both, alongside the legitimate reason of Fervo Energy’s progress.

06Claims worth citing

All figures as stated on 2025-10-16. Lead times, deployment figures and the level of hype itself all move quickly, so treat these as a snapshot.

Roughly 12,000 businesses in the Netherlands are waiting for an electricity connection, and ASML cannot build an extra factory because it cannot get a connection for the next few years. Rathi
The Netherlands has the highest per-capita rooftop solar in the world, and Rathi says it is causing real problems on local transmission and distribution networks built up over the last hundred years with much older equipment still in service. Rathi
Housing projects held up for 18 months waiting on a transformer; transformer manufacturers want something like 20 years of demand certainty before expanding capacity. Rathi
Generating capacity will not be the rate limiter on load growth at least through 2030, because there is enough capacity in enough places to serve the next few years of load growth. Kann
Gas turbines have roughly a year’s waiting list for the unit you want. This is Rathi’s phrasing in a joke comparing turbines to a sought-after toy, not a formal lead-time estimate, and it sits oddly against both speakers’ description of turbine queues as the longest in the sector. Rathi
Three manufacturers control 70 to 75% of the gas turbine market, an effective oligopoly. Rathi names them as Siemens, GE and Mitsubishi, and adds that China has tried and largely failed to build a gas turbine industry because the work is specialized, the supply chains are tight and the same parts feed jet engines. Kann for the share, Rathi for the names
The only countries with functioning civilian small modular reactors are Russia and China, which are also the countries building nuclear both domestically and abroad. Kann adds South Korea and its APR-1400, which Rathi accepts. Rathi, with Kann’s addition
The furthest-along Western reactor designs are Westinghouse’s AP1000 and possibly GE Hitachi’s BWRX-300, a 300-megawatt SMR. Kann
Fervo Energy was hoping to have the first 100 megawatts of its first big project online sometime in 2026. Kann
Bloomberg NEF projections show sodium-ion batteries taking only a sliver of the market even after 2030. Rathi’s sentence is garbled in the transcript (“even after 2030 years”), so the exact horizon he meant is unclear. Bloomberg NEF, cited by Rathi
In an interview Rathi’s programme conducted, NextEra’s chief executive said that if gas turbines are going to be that expensive, it makes solar and batteries much more attractive and they will build those instead. This is Rathi’s paraphrase from memory, not a quotation. NextEra CEO, cited by Rathi

07Where it’s contested

Co-location is the one place a verdict visibly moves, and the host’s restatement gets it wrong. Rathi first calls co-location hyped, on the grounds that you cannot get a gas turbine anyway. Kann restates that as “hyped just right, because the supply chain is the rate limiter,” and Rathi does not confirm it. Several turns later Rathi says co-location is “a little too hyped,” adding that some will get it and most will not, which is close to Kann’s overhyped verdict. Use the later statement, not the restatement.
Transformers produce the sharpest disagreement, and it is about whose queue is being measured rather than about any fact. Kann says overhyped as a bottleneck to meeting load growth; Rathi says underhyped for everyone outside the data center world. Both accept the other’s framing.
Sodium-ion. Rathi says overhyped, on Bloomberg NEF projections. Kann says hyped about right and adds that he does not think it is very hyped in the first place, pointing to CATL and other large Chinese manufacturers investing and actually deploying. He explicitly does not know whether it becomes a major chemistry shift like lithium iron phosphate or stays a niche lower-energy-density product.
Venture capital. Rathi calls the model overhyped for scaling climate solutions. Kann defends it on impact per dollar rather than on volume, and frames his own claim as historical, saying explicitly that it is not a foregone conclusion the US keeps its position going forward. Rathi’s sharper version of the worry is that the zero-to-one function depends on US basic science funding and talent, which he says is already declining under the current administration.
Kann abstains on the Paris Agreement, saying he does not hear about or think about it, essentially at all, and that he cannot say whether it should be more hyped. Treat the segment as Rathi’s answer with Kann’s abstention, not as a joint verdict.
Nuclear market structure is an open disagreement about the endgame. Kann expects consolidation toward something like the gas turbine oligopoly, and wonders whether it ends at one or two manufacturers rather than three. Rathi expects more national champions instead, because fuel transfer and which country may build where are tightly coordinated by the International Atomic Energy Agency, which blocks the market forces that consolidated gas turbines. He closes by telling Kann he is imagining a more peaceful world than the one they are heading into.
Advanced geothermal in the US is deliberately left unresolved. Kann puts it between hyped about right and slightly overhyped, notes it is very early, and says to see where it goes.
The transformer verdict carries a disclosure. Kann rates transformers overhyped as a bottleneck in the same answer in which he says he is an investor in Heron Power, a solid-state power electronics company, and expects that class of technology to revolutionize the sector. His argument does not depend on that company, but the forward-looking part of it is an investor’s expectation rather than a measured result.