It’s not inflation or recession that has blindsided traditional macroeconomists, but irrelevance. Artificial intelligence has not just transformed markets but made our dominant frameworks for understanding them obsolete.
For over a century, macroeconomics has relied on tools honed in the industrial age: GDP to measure growth, yield curves to signal recessions, and productivity metrics built around physical goods. But we now live in an economy increasingly defined by the immaterial: algorithms, platforms, synthetic data, and code.
In this world, value creation escapes the radar of traditional statistics, and policy decisions made on outdated dashboards risk steering blind. Even before the advent of generative AI, prominent economists such as Erik Brynjolfsson had proposed GDP B, a second measure of GDP to include all the real benefits and free goods or services we were getting from the digital revolution. With their antiquated tools, economists kept on complaining that they could not see any productivity uptick from the digital revolution (the same drama is now ongoing with artificial intelligence).
AI’S DYSTOPIAN INVASIONS OF PRIVACY ARE JUST BEGINNING
Consider this: S&P 500 firms today boast record average profit margins near 14%, yet that figure is an average and conceals a revolution. In the AI and software sectors, margins soar far higher, some nearing infinity in private firms with minimal headcount and maximal code leverage. Where is this margin expansion visible in our models? Nowhere. The GDP needle barely twitches. It’s like measuring a hurricane with a thermometer.
What we’re witnessing is not a speculative bubble, but a structural shift — an infrastructure revolution where the end demand, intelligence, is infinite. Data centers are the new factories for intelligence. As AI models evolve from language to vision to action — powering breakthroughs from cancer diagnostics to military logistics — the market is clearly larger than the strict generative AI sector. Compute demand isn’t cyclical. It is structurally exponential. And unlike the dot-com crash, today’s buildout is fueled not by debt-laden and low-margin telecoms, but by Big Tech companies generating vast amounts of free cash flow from their monopolistic position — Apple, Microsoft, and Nvidia — and deploying capital with strategic precision.
This has profound macroeconomic implications. First, it flips the recent inflation narrative. Despite unprecedented capital expenditure and years of central bank mistakes, we see at the end of the decade not overheating but deflationary drag. Why? Because AI boosts efficiency faster than wages can rise. Labor’s bargaining power remains weak, demographic aging continues, and most citizens are locked out of capital gains. The result is asset-rich growth and wage-poor stagnation: booming markets, limping lives.
Second, the role of government is rapidly changing. OpenAI was recently adamant that, given the size of the sector, a federal backstop was not inconceivable. Financial markets turned into video games after COVID-19, and everyone nowadays expects the Federal Reserve to be the lender of last resort to risky assets, which need funding if we want the AI revolution to continue. AI overheating is sort of blackmailing regulators and politicians.
Third, politics reframes AI as critical infrastructure. With initiatives like the U.S. “Genesis” mission — melding public data, private models, and national labs — AI is locked into a strategic rivalry with China. That makes regulation perilous: you cannot unravel what has become national security. And that draws power and energy policy into its orbit too.
So what survives in this new macro order? A few things. Old economic indicators are relics. Most corporate moats will erode under Schumpeterian AI-driven disruption. But scarcity still matters, gold, spiritualities and religion, and probably a new role for Bitcoin — a finite digital asset with a growing social consensus — which is emerging as the native store of value for a dematerialized economy.
AI IS REDEFINING NATIONAL POWER AND SOVEREIGNTY
Macroeconomics needs reckoning. It must abandon its fixation on aggregates that hide the frontier and instead embrace a probabilistic, market-derived epistemology. In a world where software eats data and data eats everything else, economic insight will no longer come from quarterly reports—but from real-time flows of code, compute, and consensus.
This isn’t the end of macroeconomics. But it is the end of macro as we know it.
Sebastien Laye is an economist and AI entrepreneur.