Most economic forecasts you hear come from people with something to sell. Erik Britton is a rare exception. He spent his early career inside the Bank of England, helped lead Oxford Economics, and now runs Fathom Consulting in London, where governments, global asset managers, and pension funds come to him for analysis that sits at the intersection of economics, geopolitics, and markets.
On this episode of Unscripted, I asked him the question every B2B leader is quietly asking: is the AI investment cycle real, and what happens to my plan if it isn't?
His answer was not hedged.
Britton framed the macro picture as a binary, and he did not leave room for a comfortable middle.
"It's not railways, it's not electrification, it's not the steam engine, it's not the jet engine, it's not the computer. It's something far, far bigger than that. Or it's a bubble. So I'm going to go with it's a bubble."
His reasoning is arithmetic, not sentiment. The returns that current capex levels are predicated on would require something like a 10 to 15 percent increase in US GDP, with all of that growth accruing to the AI sector. If that does not materialize, a large share of today's investment turns into bad assets.
Note what he is not saying. He is not saying AI produces nothing. He walked through the standard hype cycle: enormous excitement, unrealistic earnings expectations, a capex surge, a correction, and then a new baseline above where things started, reflecting the productivity gains that were genuine all along. Dot-com did this. Electrification did this. Railways did this. People made money, then lost money, and the economy ended up better off than it would have been.
The question for anyone building a plan is not whether AI is real. It is where you are standing when the correction arrives.
Britton draws a distinction that matters more than the bubble call itself.
There are two kinds of asset bubbles. The dot-com variety, where a lot of people made and lost a lot of money but very little debt was attached, produced a US recession so small it barely registered. The 2008 variety, where the inflated assets were supported by debt, took the financial system with it.
He believes this one has debt in it. "There's hidden debt, I think, in the structure of the circular financing of that whole ecosystem, especially in the US, but elsewhere too."
He places the exposure not in banks directly but in shadow banks, private equity, and venture capital, where it will eventually find its way back to the banking sector when the cycle turns.
The pre-2008 parallel he keeps returning to is the logic failure, not the instruments. Banks held that the only way their mortgage-backed portfolios could fail was if US house prices fell nationally, which had never happened. "We're creating precisely the environment in which it will happen for the first time, because you're cranking it up so much predicated on that expectation." House prices had fallen nationally in plenty of other countries. Nobody looked.
Today's version: these valuations assume nothing goes wrong to disturb the growth path. "I can tell you one thing for sure, something will go wrong. There will be a cycle, and the prices can't live with anything like that."
He also offered a historical note worth keeping in mind the next time someone points to the current market leaders as proof. Royal Bank of Scotland was the biggest bank in the world around 2007, bigger than any US bank, and bought ABN Amro at the peak. A year later you could have bought the entire UK banking sector, RBS included, for what they paid for that one acquisition, and had change left over.
Here is the part of the conversation that should reshape how revenue leaders think about team design.
Every prior technology wave displaced workers at the bottom of the income and skills distribution. The Luddites smashing weaving machines were right about their own jobs. But the next generation trained for higher-productivity work, and that cycle repeating is roughly how economies advance.
Britton, citing work by Nobel laureate Daron Acemoglu, thinks this one runs in reverse. "This technological advance is not coming after the lowest income, the lowest skilled people. It's coming after cognitive functions, which are at the higher end of the income and the skills distribution."
I pushed on the consequence that keeps coming up in client conversations: the apprenticeship problem. The analyst work, the first-draft work, the entry-level grind is how people used to earn the judgment that makes a senior practitioner valuable. Remove those rungs and you have a structural question nobody has answered. How does someone reach a senior level of experience without the apprenticeship being available?
Britton's answer was blunt. "Progressively further and further up the cognitive chain will be displaced if this is how it plays."
For marketing and revenue organizations, that is not a 2035 problem. It is a staffing model question for the next planning cycle.
Britton is not a skeptic by temperament. He is a skeptic by method, and his workflow is instructive.
He uses Claude and ChatGPT regularly. He has not embedded either in his company's workflow, for two reasons: privacy and proprietary information exposure, and fidelity of facts. For a firm whose product is rigorous analysis of data, the second one is disqualifying. "My experience thus far with AIs is very sketchy on those points. They present facts with a great deal of confidence, and it turns out half of them are not true on closer analysis."
What he does do is worth copying regardless of your industry. When a client brings a problem, he first works out how he and his team would frame it, what ideas they would bring, how they would approach it. Only then does he ask AI what it thinks, and mostly to surface existing literature. "I wouldn't at the moment rely on it to furnish the initial ideas."
Human framing first, machine assistance second. That sequence protects the thing a professional services firm actually sells.
I asked Britton how a central bank should classify AI: supply shock, demand shock, or financial stability risk.
All three, simultaneously.
It is a positive supply shock. It carries financial risk because valuations are far ahead of the genuine productivity component. And it produces a demand shock as workers are displaced and again as a correction puts more people out of work for ordinary cyclical reasons.
A positive supply shock paired with a negative demand shock is the hardest combination a central banker can face. His expectation: rates cut to the floor, potentially to zero or slightly below.
For CFOs and CEOs modeling the next two years, that is the countervailing force worth noting. The downside scenario comes bundled with very cheap money and supportive fiscal policy.
I asked what he would tell a B2B CEO or a sales and marketing leader trying to build a 2027 plan with this much uncertainty.
"Prepare yourself for a likely event, which is the bubble bursts. If you're considering a 2027 launch, get ready, because you're about to get punched in the guts, with quite a high likelihood."
Then he turned it around, and this is the line I would put on the wall.
"That doesn't mean you shouldn't start. Sometimes those are the best starts. There's going to be another side. There'll be an end to that process, and where do you want to be at the end of it? If you can survive and live through that period and come out the other side, that's a really good space to be in. If you can live there, then you can definitely live and thrive in the upside when it comes."
That is a planning posture, not a prediction. Build assuming conditions tighten. Protect the ability to keep operating through a hard year. Be positioned for the recovery rather than exiting before it.
Early in the conversation I raised the common complaint that economists are always wrong, the same thing people say about weather forecasters. His response reframes the whole discipline.
"You think you're doing without a forecast. You're not. You have a forecast when you step out of the door in the morning that the sky isn't going to fall on your head. You have a model in your mind, and you're generating forecasts based on that for absolutely everything you do. The question is not are you using one. The question is how good is it."
Every revenue plan is a forecast. Every pipeline coverage ratio is a model. The useful question is not whether to forecast under uncertainty. It is whether your model is rigorous enough to be worth acting on, and whether you have built in the room to be wrong.
Erik Britton of Fathom Consulting argues it is, on the grounds that current capex levels imply returns requiring roughly a 10 to 15 percent increase in US GDP accruing to the AI sector. He expects a correction followed by a new baseline above the pre-hype level, reflecting the genuine productivity gains, which is the pattern that followed dot-com, electrification, and railways.
Because of debt. The dot-com unwind destroyed equity value but carried little leverage, producing a very mild US recession. Britton believes the current cycle involves hidden debt in circular financing structures, concentrated in shadow banks, private equity, and venture capital, which puts the financial system at greater risk.
Previous technology waves displaced workers at the bottom of the income and skills distribution. AI targets cognitive functions at the higher end, including the entry-level roles within professional careers that served as the apprenticeship path to senior expertise.
Plan for a recessionary environment with a high likelihood of disruption, while recognizing that such periods typically bring very cheap money and supportive fiscal policy. Britton's guidance is to build for survival through the downturn rather than delaying, because the position you hold coming out the other side is where the advantage is.
Britton's approach is human framing first, AI second. He develops his own and his team's view of a problem before consulting AI, and cites factual fidelity and proprietary data exposure as the reasons he has not embedded it in Fathom's core analytical workflow.
Listen to the full episode of Unscripted with Jeff Pedowitz featuring Erik Britton, managing director and CEO of Fathom Consulting.