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.
[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?
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.
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] |
| Global wire-transmissible wages (author's extrapolation) | ~$18–25 trillion[54] |
| Global labor income (labor's share of world GDP) | ~$60–62 trillion[24][25] |
So now there are three worlds we can find ourselves in, and you can read them straight off the jobs data.
More layoffs. Substitution, the simple math: fire people, hire tokens. Wages avoided, visibly.
Hiring frozen. The quiet version of the same thing. Uber spends a dollar on tokens to stand up a bike-courier business. The token bill still gets paid (customers fund it, the way passengers fund jet fuel), but the dispatch, support, and routing jobs that line would have needed in 2019 never get posted. And the freeze isn't a knife-edge between firing and hiring, because tokens take the work that scales with volume and leave humans the work that scales with exceptions, and that's coverage, a step, not a dial: two pilots per cockpit however good the autopilot gets, and a third adds nothing. The veteran is worth keeping, the new hire adds roughly zero, and attrition shrinks a frozen company 10 to 15% a year without a single headline. A layoff on layaway: it's how teller ranks thinned after mobile banking[46], it's what Klarna ran in public[47][48], and it's why the canary lives at the bottom rung, since the routine work that used to justify, and train, the junior hire is exactly what the tokens took[52]. Either way, and this is the spine of the piece, the price a token can charge is anchored to the cost of the human who would have done the task. Wages paid, or wages avoided.
More hiring. Genuine expansion. This is the scenario everyone is hoping for: companies make new revenue on top of the token bill, hire more people, and grow faster, with the machines and the humans compounding each other. So price the hope. By 2031 the bill is about $2.5 trillion a year, and a buyer only comes out ahead if the tokens generate more than they cost, so the world's companies would need to mint at least $2.5 trillion of genuinely new output a year, plus a margin for themselves. That's a little over 2% of world GDP[25] in brand-new production, an economy roughly the size of Italy[29], materializing and then sticking around. And it has to show up fast: the chips need paying back inside a three-to-six-year window, while the new-revenue stories run on longer clocks; a new drug can still take roughly a decade to develop and approve[30]. The growth is possible. The depreciation is scheduled.
One honest exception to the wage anchor: work with no human alternative at all, an agent grinding through a million-step problem overnight, something no team could do at any price. There, the anchor is what someone will pay, not what a person costs. That category is real. Nobody has measured it; my judgment is that today it's small.
What decides which world: three gates #
Which world you get isn't a mystery. A century and a half of automation, and the task framework Daron Acemoglu and Pascual Restrepo built to explain it[43], says it comes down to three gates, in order.
Gate one: the size of the dividend. Is the saving big or small?
Small savings displace the worker without moving the price. Acemoglu's name for this is "so-so automation"[43], and his example is self-checkout: the cashiers are gone, the groceries cost the same, and now you do the scanning for free. Phone-tree customer service, same story: the operators left, the service got worse, the savings were marginal. The closest dress rehearsal for token spending is the enterprise RPA wave of 2015 to 2020: heavily hyped, big license spend, and EY itself warned that 30 to 50% of initial projects were failing[61]. Savings that never showed up anywhere a customer could see.
Big savings change everything downstream. The assembly line took the Model T from about $850 to about $290[58], turned the car from a luxury into a universal product, and grew motor-vehicle employment from roughly 85,000 wage earners to about 400,000 in fifteen years[58]. Containerization cut the cost of a ship from $5.83 a ton to about 16 cents and is credited with boosting trade among industrialized countries by roughly 500% over the following fifteen years[59]. Spreadsheets erased roughly 400,000 bookkeeping and accounting-clerk jobs and added roughly 600,000 accounting jobs[60]. ATMs cut tellers per branch from about 21 to 13, which made branches cheaper to open, so banks opened 43% more of them, and teller employment rose for three decades, until mobile banking automated nearly the whole job and the decline finally came[46].
Notice the tell that separates the two lists: in every big-dividend case, a price collapsed. In the so-so cases, it didn't.
Gate two: demand elasticity. A big dividend isn't enough; people have to want more of the thing when it gets cheap. The warning here is the tractor. Agriculture's dividend was enormous, nothing so-so about it. But people can only eat so much. Farm work went from about 41% of the American workforce in 1900 to under 2% anyway[56], and the gains funded other sectors instead of more farming. The open question for cognitive work is which kind it is. A lot of it is intermediate demand (support tickets, compliance reviews, invoice processing), things firms want less of per unit of business, not more. That's agriculture-shaped. Some of it (creation, entertainment, software people actually want) might be genuinely elastic. Nobody knows the split, and the split is most of the answer.
Gate three: reinstatement. Even when the first two gates open, the new human work has to actually appear. Historically it has: about 60% of the jobs Americans do today are in occupations that didn't exist in 1940[45]. But it's the slowest gate, and the gap between displacement and reinstatement is where the pain lives.
One more number before the receipts. In 2024, Acemoglu priced the measurable path, the tasks AI can already do and the savings on each, at a total-factor-productivity gain of no more than 0.66% over ten years, on the order of 1% of GDP[44]. He explicitly declined to price new products and new tasks. So the leading academic estimate of the first two worlds comes to about one percent of GDP, and the entire bull case lives behind gates two and three, in the world nobody can measure yet.
The dividend smoke test #
A big dividend can't hide. When the price of a core input collapses, it leaves smoke everywhere: in prices, in margins, in behavior, eventually in the macro numbers. Containerization showed up in costs immediately and in trade volumes within a decade. So instead of arguing about the future, inventory the smoke.
Price smoke: none. Software isn't getting cheaper for the buyer. Neither is legal work, consulting, or anything support-heavy. Three years into the boom, try to name one price a customer pays that has collapsed because of tokens.
Corporate smoke: cost numbers only, and small ones. JPMorgan's roughly $600 million of efficiencies[40] is about a third of one percent of its $185.6 billion of 2025 revenue[57], a so-so-sized number at big-dividend prices. Morgan Stanley finds about a quarter of the S&P 500[33] (roughly 40% of the companies it counts as real adopters[34]) reporting some measurable benefit, and "at least one measurable benefit" is a low bar read off earnings calls by AI. What doesn't exist anywhere: a large company crediting tokens with a point of revenue growth.
Volume without revenue. Sensor Tower estimates about 235,800 App Store submissions in Q1 2026, up 84% from a year earlier[37]. Shots on goal, everyone credits the coding tools, and there's no matching consumer spend on the other side. Volume smoke without price smoke or revenue smoke is exactly the so-so pattern.
Behavioral smoke points the wrong way. Uber capped AI coding tools at $1,500 per employee per month[31] after teams burned the annual budget in four months[32]. You don't cap a gusher. Klarna announced its AI assistant was doing the work of 700 support agents[47], then spent 2025 recruiting humans back into customer service after its CEO admitted cost had been "a too predominant evaluation factor" and quality had suffered[48]. And a preliminary MIT-affiliated report[36] found roughly 95% of corporate AI pilots invisible in the P&L. Early-stage work (52 interviews, 153 surveys, never peer-reviewed), so hold it loosely. But it rhymes.
Macro smoke: none yet. The cleanest reading comes from the San Francisco Fed: labor productivity has run hot since 2023, but total factor productivity, the series a genuine technology dividend is supposed to show up in, has barely moved, and their model puts just a 21% probability on a high-growth regime by the TFP data[49]. Their interpretation is close to this essay's: the gains so far look like better tools for workers, not a broader transformation. An NBER survey of nearly 6,000 executives[35] agrees from the inside: 89% saw no productivity impact from AI over the past three years, and over 90% saw no effect on headcount. The honest caveat: every general-purpose technology has lagged in the statistics, and electricity took decades. The lag excuses the absence. For now.
Genuine smoke, in exactly two narrow places. Translation is the cleanest hit in the labor data: a one-point rise in machine-translation use cut translator employment growth by about 0.7 points across US cities, roughly 28,000 positions that never got created[51], and 36% of translators report losing work to generative AI, with 43% reporting falling income[50]. (Rates are under pressure too, but nobody has published clean price data, so I'll leave that claim on the table.) Coding is the other: the velocity gain is real, and the monetization of that velocity is unproven. See the App Store bullet above. The question that decides everything is whether these are the first two verticals of fifty, or the only two.
And the seller layer is billowing. Google Cloud grew 82% last quarter, to $24.8 billion, with $8.8 billion of operating income at a 36% margin[8]. The segment bundles plenty of non-AI business, but somebody is unambiguously getting paid. Thick smoke at the shovel-sellers and none at the diggers isn't evidence against this essay. It's the first ending, already in progress.
One more thing about why the smoke is thin. Layoffs make headlines; frozen requisitions make nothing. The so-so world doesn't arrive as an event, it arrives as an absence. That's why "no effect" surveys and a so-so trend are perfectly compatible, and it's why entry-level hiring is the canary: early-career workers in the most AI-exposed occupations have already seen 16% relative employment declines while their experienced colleagues held steady[52]. Nobody announces the job that never gets posted.
Three endings, in hiring terms #
By this point, token spending comes in two shapes, and they don't behave the same. Some of it is cost of goods, tokens baked into a product that sells, like the bike business. That spending is durable. Nobody cuts the ingredients of a product that's selling: the airline industry has famously failed to earn its cost of capital across its history[38] (it still doesn't[39]), and the fuel bill got paid through all of it. The rest is insurance: defensive spending with no product attached, bought so nobody can say you fell behind. That's the fragile part.
So: three endings, and you can tell them apart from the jobs data alone.
One: the race continues. The cost-of-goods spending compounds, fear keeps the insurance renewed, and the compute layer gets paid out of the adopters' hides. In hiring terms: the freeze deepens. No bloodbath, no headline, just a labor market that quietly stops absorbing new people, one unposted requisition at a time.
Two: the insurance gets cancelled. Enough CFOs ask "where's the return?" in the same quarter. What synchronizes them? The usual things: a rate spike, an earnings recession, one bad quarter that makes every board ask why the AI line doubled. Insurance is an expense you cut the moment money gets tight, and this is the most expensive insurance ever sold. The cost-of-goods spending survives that moment; the insurance doesn't. My read of the smoke above is that most of today's spending is still insurance-shaped, and if that's right when the trigger comes, the freeze thaws, and a large piece of that $7.6 trillion is stranded, never earned back by anyone.
Three: the so-so world arrives. The dividend turns out real but small: self-checkout at civilizational scale. The layoffs and freezes happen anyway, because the token bill has to be funded, but prices don't fall and the pie doesn't grow. The wage pool shrinks with nothing handed back to the customer, who is, of course, the same person who lost the wages. This is the ending nobody's modeling: buyers don't win, workers don't win, and the customer base the whole machine sells into slowly erodes underneath it.
Notice what all three endings share: the companies buying the tokens don't come out ahead. In a competitive market, lasting extra returns for adopters trend toward zero. The only live question is how much the compute layer keeps before its own prices get competed down too.
Yes, there are exceptions. A company with data nobody else has, locked-in distribution, or a regulator keeping rivals out can hold onto its gains. The question was never whether exceptions exist. It's whether they're anywhere close to big enough to cover a $7.6 trillion tab.
Sources & References #
- Goldman Sachs, "Tracking Trillions: The Assumptions Shaping the Scale of the AI Build-Out," Goldman Sachs Insights, 2026. Link - Microsoft, FY2026 Q4 earnings materials and call, Microsoft Investor Relations, 2026. Link - Meta Platforms, "Meta Reports Second Quarter 2026 Results," Meta Investor Relations, 2026. Link - Associated Press, coverage of Amazon's capital-expenditure guidance, AP News, 2026. Link - Investing.com, "Google Quarterly Cloud Revenue Growth Beats Expectations" (Alphabet capex guidance), 2026. Link - Oracle Corporation, "Oracle Announces Record Q4 and FY2026 Results Driven by Cloud Infrastructure & Cloud Applications," Oracle Investor Relations, 2026. Link - Benzinga, "Full Transcript: Oracle Q4 2026 Earnings Call," 2026.
[Link](https://cdn2.benzinga.com/insights/news/26/06/53131763/full-transcript-oracle-q4-2026-earnings-call) - Alphabet Inc., Q2 2026 Earnings Release, 2026.
[Link](https://s206.q4cdn.com/479360582/files/doc_financials/2026/q2/2026q2-alphabet-earnings-release.pdf) - Bank of England, Financial Stability Report, December 2025.
[Link](https://www.bankofengland.co.uk/financial-stability-report/2025/december-2025) - Bank of England, Financial Stability Report, July 2026.
[Link](https://www.bankofengland.co.uk/financial-stability-report/2026/july-2026) - Alphabet Inc., Investor Relations FAQ (server depreciation schedule).
[Link](https://alphabet2025ir.q4web.com/investor/faqs-and-general-information/) - Princeton CITP Blog, "The Lifespan of AI Chips: The $300 Billion Question," October 15, 2025.
Link - Microsoft, "The Next Phase of the Microsoft-OpenAI Partnership," Official Microsoft Blog, April 27, 2026. Link - Federal Trade Commission, "FTC Issues Staff Report on AI Partnerships & Investments Study," press release, January 2025. Link - Anthropic, "Anthropic and Amazon Expand Partnership with Trainium," Anthropic News. Link - Federal Trade Commission, 6(b) staff report on AI partnerships and investments (redacted). Link - Anthropic, "Expanding Our Use of Google Cloud TPUs and Services," Anthropic News. Link - Nvidia Corporation, Schedule 13G filing disclosing stake in CoreWeave, SEC EDGAR. Link - Nvidia Corporation, press-release exhibit on additional CoreWeave investment, SEC EDGAR. Link - Investing.com, "CoreWeave, Nvidia Sign New $6.3 Billion Deal for Cloud Computing Capacity," 2026. Link - MIT Center for Transportation & Logistics, "Telecom, Cisco and Lucent" (vendor-financing case study). Link - Sensor Tower, "State of AI 2026 Report: Global Time Spent on Generative AI Apps Projected to More Than Double Year-Over-Year." Link - U.S. Bureau of Economic Analysis, "National Economic Accounts Annual Update," Survey of Current Business, November 2025.
[Link](https://apps.bea.gov/scb/issues/2025/11-november/1125-nea-annual-update.htm) - UN Statistics Division, "Extended Report 2025: SDG Goal 10."
[Link](https://unstats.un.org/sdgs/report/2025/extended-report/Extended-Report-2025_Goal-10.pdf) - Federal Reserve Bank of St. Louis, FRED series "World GDP" (NYGDPMKTPCDWLD).
Link - World Intellectual Property Organization, "Global Software Spending," Global Innovation Index blog, 2025. Link - Gartner, "Gartner Forecasts Worldwide IT Spending to Grow 13.5% in 2026, Totaling $6.31 Trillion," April 22, 2026. Link - "Payrolls to Prompts," arXiv preprint, 2026. Link - Federal Reserve Bank of St. Louis, FRED series "Italy GDP" (MKTGDPITA646NWDB). Link - U.S. Food and Drug Administration, "Drug Development and Review Definitions." Link - Los Angeles Times, "Uber Caps Staff Use of AI Coding Tools After Blowing Its Budget," June 2, 2026. Link - TechCrunch, "Uber Caps Employee AI Spending After Blowing Through Budget in Four Months," June 2, 2026. Link - Morgan Stanley, "Economic Signals Emerging from Founders Institute 2026," Morgan Stanley Insights. Link - The Corner, "25% of S&P Companies Already Reporting Measurable Benefits from AI." Link - NBER Working Paper No. 34836, executive survey on AI adoption and productivity. Link - Project NANDA (MIT), "The State of AI in Business 2025" (archived). Link - 9to5Mac, "App Store Sees 84% Surge in New Apps as AI Coding Tools Take Off," April 6, 2026.
[Link](https://9to5mac.com/2026/04/06/app-store-sees-84-surge-in-new-apps-as-ai-coding-tools-take-off/) - McKinsey & Company, "The Six Secrets of Profitable Airlines."
[Link](https://www.mckinsey.com/industries/travel/our-insights/the-six-secrets-of-profitable-airlines) - International Air Transport Association, press release, December 9, 2025.
[Link](https://www.iata.org/en/pressroom/2025-releases/2025-12-09-01/) - JPMorgan Chase & Co., 2026 Company Update, full event transcript.
Link - GuruFocus, "Meta Platforms Free Cash Flow," trailing-twelve-month data, accessed August 2026. Link - Amazon.com, Inc., "Amazon.com Announces Second Quarter Results," Q2 2026 earnings release, SEC EDGAR. Link - Daron Acemoglu and Pascual Restrepo, "Automation and New Tasks: How Technology Displaces and Reinstates Labor," Journal of Economic Perspectives, 33(2), 2019. Link - Daron Acemoglu, "The Simple Macroeconomics of AI," NBER Working Paper No. 32487, 2024. Link - David Autor, Caroline Chin, Anna Salomons, and Bryan Seegmiller, "New Frontiers: The Origins and Content of New Work, 1940–2018," The Quarterly Journal of Economics, 139(3), 2024. Link - James Bessen, "Toil and Technology," IMF Finance & Development, March 2015. Link - Klarna, "Klarna AI Assistant Handles Two-Thirds of Customer Service Chats in Its First Month," press release, February 2024. Link - Charles Daly, "Klarna Turns From AI to Real Person Customer Service," Bloomberg, May 2025. Link - Hamza Abdelrahman and Andrew Foerster, "Have We Entered an Era of High Productivity Growth?," FRBSF Economic Letter 2026-14, Federal Reserve Bank of San Francisco, 2026. Link - The Bookseller, "A Third of Translators Report Losing Work to Generative AI Systems, SoA Survey Reveals," 2024. Link - Carl Benedikt Frey and Pedro Llanos-Paredes, "Lost in Translation: Artificial Intelligence and the Demand for Foreign Language Skills," Oxford Martin School, 2025. Link - Erik Brynjolfsson, Bharat Chandar, and Ruyu Chen, "Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of Artificial Intelligence," Stanford Digital Economy Lab, 2025. Link - U.S. Bureau of Labor Statistics, "Occupational Employment and Wages, May 2025," Table 1, 2026. Link - Jonathan I. Dingel and Brent Neiman, "How Many Jobs Can Be Done at Home?," Journal of Public Economics, 189, 2020. Link - Tyna Eloundou, Sam Manning, Pamela Mishkin, and Daniel Rock, "GPTs are GPTs: Labor Market Impact Potential of LLMs," Science, 2024. Link - Carolyn Dimitri, Anne Effland, and Neilson Conklin, "The 20th Century Transformation of U.S. Agriculture and Farm Policy," USDA ERS Economic Information Bulletin No. 3, 2005. Link - JPMorgan Chase & Co., "Fourth-Quarter and Full-Year 2025 Financial Results," 2026. Link - Gene Smiley, "The U.S. Economy in the 1920s," EH.net Encyclopedia of Economic History, 2004. Link - Daniel M. Bernhofen, Zouheir El-Sahli, and Richard Kneller, "Estimating the Effects of the Container Revolution on World Trade," Journal of International Economics, 2016. Link - David Kestenbaum and Jacob Goldstein, "Episode 606: Spreadsheets!," NPR Planet Money, 2015. Link - EY, "Get Ready for Robots: Why Planning Makes the Difference Between Success and Disappointment," Ernst & Young, 2016. Link