McGrann argues that if you subtract AI-related capex from GDP, the rest of the economy is functionally dead. He points to hyperscaler capex exceeding the dot-com peak as a share of S&P 500 capex, with Nvidia, Microsoft, Google, Meta, Amazon, and Oracle driving most index gains while mid/small-cap revenue is flat-to-negative and SMB hiring has stalled.
By submitting the essay and driving it to 1,127 points, the HN community signal-boosted the thesis that the headline economy and the practitioner economy have decoupled. The submission's reception reflects engineers' lived experience of the worst job market since 2009 coexisting with record markets.
The editorial frames McGrann's contribution as the move from treating the markets-vs-jobs split as a paradox to treating it as an accounting artifact. It notes this is the third or fourth HN front-pager in six weeks attempting this reconciliation, with wage growth for senior ICs at non-AI shops essentially zero year-over-year.
The editorial highlights that 2025 hyperscaler capex tracking north of $400B shows up simultaneously as GDP investment, Nvidia revenue, and depreciation that will haunt these companies' future earnings. The framing implies the current 'growth' is a forward liability disguised as present-day output.
Owen McGrann's essay 'The Dead Economy Theory' hit 1,127 points on Hacker News this week, the latest entry in a growing genre: practitioner-written macro pieces that argue the official numbers no longer describe the economy developers actually work in. McGrann's core claim is brutally simple: take AI-related capex out of GDP and the rest of the economy is, functionally, dead.
The specifics he marshals are familiar to anyone who has been reading earnings calls instead of press releases. Hyperscaler capital expenditure is now running at a pace that, as a share of S&P 500 capex, exceeds the dot-com peak. Nvidia, Microsoft, Google, Meta, Amazon, and Oracle account for a wildly disproportionate share of recent index gains. Outside that cluster, revenue growth at mid-cap and small-cap firms is flat-to-negative in real terms. SMB hiring has stalled. Wage growth in software, after a decade of compounding, is essentially zero year-over-year for senior ICs at most non-AI shops.
The essay's rhetorical move is to treat GDP not as a measure of economic health but as a leaderboard distorted by one team's spending. If a single category — datacenter build-out — is responsible for the majority of marginal growth, then 'the economy is growing' and 'the economy is dying' can both be true, depending on whether you're inside that category or outside it.
This is the third or fourth time in six weeks that an HN front-pager has tried to reconcile two facts that obviously contradict each other: markets at all-time highs, and a job market that senior engineers describe as the worst they've seen since 2009. McGrann's contribution is to stop treating that as a paradox and start treating it as an accounting artifact.
The math is worth being precise about. Hyperscaler capex for 2025 is tracking north of $400B. That spending shows up in GDP as investment, in equity markets as Nvidia revenue, and in earnings as depreciation that will haunt these companies for the next decade. What it does not show up as is payroll growth, SMB revenue, or consumer spending power — the three things that historically translated 'GDP up' into 'my friends are getting jobs.' The transmission belt between corporate spending and the rest of the economy has been replaced by a closed loop: cloud providers buy GPUs, GPU revenue inflates one stock, that stock's gains accrue to a narrow band of investors, and the broader labor market sees none of it.
This matters for devs in a way that goes beyond schadenfreude about macro. If you're trying to read your own career through the lens of 'tech is booming,' you will misprice every decision in front of you. The headline says hiring is strong; your recruiter pipeline says it isn't. The headline says comp is rising; your annual review says it isn't. The headline says AI is creating jobs; LinkedIn says the jobs being created are at six companies. The dissonance isn't in your head — it's in the index.
There's also a circuit-breaker risk that doesn't get enough airtime. If hyperscaler capex flatlines for even two quarters — because token revenue disappoints, because Lombardy-style datacenter restrictions spread, because one fund finally calls the bubble — the GDP print goes negative and the labor market goes from frozen to actively contracting. The 'dead economy' under the AI layer isn't dormant; it's load-bearing weight it can't actually carry. The capex spend is the thing keeping the appearance of growth alive, and that spend is built on a token-revenue forecast that, as we noted earlier this week, requires roughly $1T/year just to break even on the announced builds.
The community response on HN was telling. The top comment threads weren't arguing with McGrann's numbers — they were arguing about which sectors he'd undercounted. Construction outside datacenters: dead. Commercial real estate: dead. Consumer discretionary ex-AI-adjacent: dead. Even biotech, traditionally the second-largest VC magnet, has seen funding collapse. Whatever you want to call this, it isn't a normal expansion.
For practitioners, the dead economy theory has three concrete implications worth internalizing.
First, be skeptical of any career advice premised on macro tailwinds. The advice 'just switch jobs, the market is hot' assumes a market that hasn't existed for eighteen months outside of AI labs. If you're not in the capex-recipient cluster — meaning Nvidia, the big four clouds, a top-five lab, or a vendor selling directly into datacenter build-out — your local labor market looks like 2009, regardless of what the S&P prints. Optimize accordingly. Stay where you are unless the new role is materially better, not just nominally newer.
Second, be honest about what your employer is actually selling. If your company's revenue depends on SMB IT budgets, mid-market SaaS spend, or consumer discretionary, the macro picture is worse than your CEO is admitting on the all-hands. The deck shows AI tailwinds; the pipeline shows deal slippage. Build for that. Cut scope, lengthen runway assumptions, and treat any 'we're hiring again next quarter' assurance with the skepticism it has earned.
Third, the architecture choices you make now should price in a non-zero probability that the capex spigot tightens. If you've built your stack around assumed-cheap, assumed-abundant frontier-model inference, you've made a bet on the same cluster of companies that the dead economy theory says is propping up the entire index. Have a fallback. Local models, smaller models, fewer model calls, cached results — all of these look like cost optimizations today and look like survival adaptations if Q3 capex guidance comes in soft.
The dead economy theory is not a forecast; it's a diagnosis. The forecast it implies is that one of two things resolves the contradiction: either the capex layer eventually generates the revenue to justify itself and the rest of the economy catches up, or the capex layer breaks first and takes the rest of the index down with it. There is no third option in which GDP stays elevated, AI capex slows, and the SMB economy quietly heals. McGrann's piece won't be the last HN front-pager arguing this, but it's the clearest framing of why the numbers your manager quotes and the reality your team feels have drifted so far apart.
Not many people see that the end result is that when ordinary people stop earning money, and stop buying products, all the money will be used purely for B2B transactions.Money will still exist, but people will not see hardly any of it. To break out of being just a person and start a business you wil
This article, like Citrini research's scenario before it, misses much of the economics.AI is unlikely to be as revolutionary as is presumed. It's definitely going to lead to increased productivity, and will probably render some jobs redundant, but it's unlikely to have a significant e
This article puts into words a lot of things that had been on my mind as missing in AI discourse. Most significantly, actually considering the _systemic consequences_ of the promised AI future, how it interacts with political economy, an actual critical look (instead of accepting the "Western m
It’s great this debate is happening. The people on hn are some of those that can make a real impact here. There are many parallels with the industrial revolution here but we really don’t want to repeat a over a century of misery that occurred after that era. If we don’t want the world to be run be a
Top 10 dev stories every morning at 8am UTC. AI-curated. Retro terminal HTML email.
The article doesn't seem to take his train of thought quite far enough.If AI suddenly makes it possible for a law firm to be run with a skeleton crew, then what's stopping all those people you fired from starting new law companies, where AI also does most of the work, and competing with yo