So Who's Going to Buy All These Tokens?

Every AI bear case you've read is a bet against the technology. The models are overhyped, the agents don't work, the pilots quietly die. Ed Zitron has built a whole beat on it. This piece makes the opposite bet. Assume the models work. Assume they keep getting better. Assume every demo ships. The problem isn't the technology. It's the customer's math: all this spending only makes sense if somebody buys roughly $1.2 trillion of tokens a year, climbing toward $2.5 trillion by 2031. So who's the buyer? And do they ever get their money back? Short answer: the buyer is payroll, and no. And you won't have to wait a decade of productivity statistics to check my work. It shows up first in one place: hiring. What hiring already shows is the bad version: so-so automation, jobs traded for tokens with the dividend still missing. The tab Let's try to get a grasp of the size of this whole boondoggle. Start with what's being spent. Goldman Sachs models AI capex[1] at $765 billion this year, rising to $1.6 trillion a year by 2031; that's $7.6 trillion all in. Obviously that's a projection, not a promise, but company guidance is in the same neighborhood. Microsoft[2] is guiding to about $175 billion of reported capex this year (roughly $190 billion before a lease-accounting change), Meta[3] to $130–145 billion, Amazon[4] to about $220 billion, and Alphabet[5] to $195–205 billion. That's roughly $733 billion for the four of them at the midpoints, or call it $748 billion on Microsoft's old basis. Oracle[6] runs on its own fiscal calendar, but its next-year indication works out to as much as $95 billion gross[7]. Stack them all and you're in the $830–840 billion range, with the caveat that the fiscal periods and definitions don't line up perfectly. It's not spread evenly. Oracle just finished a fiscal year spending $55.7 billion on capex against $67.4 billion of revenue[6]. That's 83 cents of every dollar it brought in. The big four are nowhere near that, but they're all at levels that would have looked insane five years ago. 2026 Capital Expenditure Guidance, by Company Capex is single-year guidance as disclosed by each company; fiscal periods and definitions vary, and ranges are shown at their stated midpoint. None of the five gives forward free-cash-flow guidance, so the second series shows what each actually reported most recently[2][6][8][41][42]: full fiscal-year totals for Microsoft (FY26, ended June 2026) and Oracle (FY26, ended May 2026), trailing twelve months through the most recent quarter for Meta, Amazon, and Alphabet. Can cash flow cover this? Not really. Alphabet, the best case, generated $53 billion of free cash flow over the past year even after $132 billion of capex[8]. Oracle just ran a fiscal year about $24 billion free-cash-flow negative[6]. Sector-wide, the Bank of England estimates the buildout needs roughly $1.5 trillion of outside money[9] under current plans, including about $800 billion from private credit (a forward-looking estimate for the whole sector, refreshed this July[10], not a claim that every checkbook is empty). Some of it is plain debt; a lot of it is leases, capacity deals, and private-credit structures. Without getting a CPA involved, let's do some quick payback math. AI chips don't earn forever. Alphabet books its servers over about six years[11], but a chip can stop earning premium rates long before the accounting says so[12]: the next generation shows up and undercuts it. And once you're buying new hardware every single year, the question stops being "when does this batch pay off" and becomes "what does the whole machine need to earn, every year." That's just annual capex divided by the margin on compute. Say the sellers keep 65 cents of each revenue dollar after the direct cost of serving it. That number is my assumption; move it if you want, I don't care. The $765 billion of capex spend needs about $1.18 trillion of revenue a year just to cover the hardware. We're talking about practice, not the game, practice. That's before salaries, buildings, interest, or a single dollar of return for anyone. At 2031's spending rate, the number is about $2.5 trillion. So who is going to spend that? The Hardware Alone Needs a Trillion-Dollar Customer Annual capex versus revenue required to cover just the hardware, assuming sellers keep 65¢ of every compute revenue dollar after direct serving costs (author's assumption; move it and the hurdle moves). So What's the Demand Story Today? It's complicated. A lot of today's "demand" is the supply side buying from itself. Microsoft's money flows to OpenAI, and OpenAI's compute runs on Microsoft's cloud[13], and the FTC found these partnerships came with requirements to spend big chunks of the investment right back on the partner's cloud[14]. Amazon has put $8 billion into Anthropic[15], Google another $2.55 billion[16], and Anthropic buys enormous amounts of compute from both[17]. Nvidia owns a piece of CoreWeave[18], added $2 billion more[19], and commits to buy up to $6.3 billion of CoreWeave's unsold capacity through 2032[20]. Real money moves, real revenue gets booked, and some of it may even reflect real demand. But entangled revenue can't prove the thing this essay needs proven: that an outside customer, spending only its own money, will pay. Three Closed Loops Behind “Demand” Each investor's cash comes back to it as a customer payment for cloud or compute: the same dollars, changing hats. This doesn't make the money fake; it just means it can't prove an outside buyer exists. invests pays for Azure compute Microsoft OpenAI $8B invests pays AWS $2.55B invests pays Google Cloud Amazon Google Anthropic equity stake + $2B buys $6.3B capacity thru '32 Nvidia CoreWeave Investment Compute / cloud purchase Microsoft, Amazon, Google, and Nvidia each invest in an AI lab or infrastructure company that turns around and spends a large share of that money buying compute or cloud capacity back from the same investor.[13][14][15][16][17][18][19][20] We've seen this movie. In the late-90s telecom bubble, Lucent extended about $8 billion of financing to its own customers, and Nortel about $3 billion[12]. Vendor financing made demand look structural when a lot of it was the sellers funding their own order books[21]. And keep the ending of that story in mind: much of the fiber was real and eventually useful. It got lit, it carried the internet, consumers won huge. The overbuild destroyed enormous amounts of investor capital anyway. Useful infrastructure and destroyed capital are not opposites. That doesn't make the labs' revenue fake. It means you have to count it carefully: money from actual outside companies and actual consumers counts. Money from your own investors, partners, and suppliers doesn't. Consumers are real, just small. AI apps pulled in over $4 billion of in-app purchases in the first half of 2026[22], a number that leaves out web subscriptions and API deals, and one that's growing fast. It's still $4 billion per half-year against a hurdle of a trillion per year. Consumers are real, just small. AI apps pulled in over $4 billion of in-app purchases in the first half of 2026[22], a number that leaves out web subscriptions and API deals, and is growing fast. It's still $4 billion per half-year against a hurdle of a trillion per year. Wages, Wages, Wages So where does a trillion a year in tokens come from? Pretend for a second you're the CFO at Uber (trust me, you're qualified). Suddenly you have a $100 million token bill. You can either cut $100 million somewhere, or plan for your top line to grow by $101 million. The explosive scenario is that now every person you hire is 10x more productive and your revenue per employee goes way up. Either way, the pool is the same one. US employee compensation runs about $15 trillion a year[23]. Worldwide, labor's share of GDP is a bit over half[24], so against a world economy of about $118 trillion[25], call it $60 to 62 trillion. But be honest about the subset: tokens don't threaten plumbers or heart surgeons, and hands-on work is most of the pool. Count only work that can travel down a wire, about 37% of US jobs, carrying about 46% of wages because desk jobs pay better[54] (an upper bound, by the authors' own description), and you get roughly $6 to 7 trillion in the US, maybe $18 to 25 trillion worldwide; that's my own wage-weighted extrapolation from their occupational classification, since remote-capable shares drop fast outside rich countries[54]. Exposure isn't all-or-nothing either: the best task-level estimates put about 19% of workers with half their tasks exposed to models and 80% with at least a tenth[55], so the addressable pool is really task-slices across many jobs, not whole occupations. And coding, the one vertical with proven big-dividend smoke, is about 5.3 million US workers earning roughly $630 billion[53]: the entire wage pool of the conquered territory is less than one year of the buildout. So the honest math: a $1 trillion token bill is about five cents of every addressable wage dollar. Payable. But it's a third the size of the headline pool, and the bulls need territory nobody has taken yet. For scale, estimates of the entire global software market run from about $700 billion[26] to $1.4 trillion a year[27], depending on what you count. And you can already see the first drops moving: one study, "Payrolls to Prompts"[28], found the companies most exposed to AI increasing their spending with model providers while cutting spending on hiring marketplaces, relative to everyone else. Tokens in, contractors out. Pool of money Annual size US computer & mathematical occupations (the conquered territory)~$0.63 trillion[53] Today's global software market$0.7–1.4 trillion[26][27] Token revenue needed to cover hardware today~$1.2 trillion Token revenue needed to cover hardware by 2031~$2.5 trillion US wages that can travel down a wire~$6–7 trillion[54] US employee compensation~$15 trillion[23]