My dad is a Lead Software Developer who has been writing code for decades, and every once in a while he forwards me something from his morning inbox when it crosses into what I’ve been talking to him about.
Lately, I’ve been telling him two things about the artificial intelligence industry. First, right after Anthropic swung back into the red—putting every single frontier AI lab at a loss this quarter—I told him that these AI companies not making money is the clearest indication that this is an infrastructure project, not a normal software business. Second, I told him that calling this an “AI bubble” is a trap, because calling it an AI bubble assumes the big investors actually treat their investment like an AI software play—when the smart money knows they are funding Q-Day infrastructure.
Then at 6:51 AM last week, The New York Times dropped its flagship morning newsletter—The Morning: A.I. by the numbers, written by Sam Sifton and Evan Gorelick—into millions of inboxes across the country. My dad saw the giant photo of an empty white podium bearing the Anthropic logo (”Code w/ Claude”) and forwarded it to me.
I opened the email expecting to just laugh at the valuation numbers. Instead, within the first three paragraphs, I watched the New York Times pull one of the cleanest, most shameless pieces of statistical propaganda I have ever seen in print.
Let’s walk through the exact text they sent out to the country that morning, because once you see how the sleight of hand works inside a single word, you can never unsee it.
The $126,700 Ghost Household
(How to Erase an American Worker in Three Moves)
The newsletter opens with Anthropic’s plan to debut on the Nasdaq at a potential $2 trillion valuation, and then tries to help the reader visualize how big $2 trillion is:
“Here’s what that number looks like, written out: $2,000,000,000,000. It’s hard to imagine. Two thousand billion dollars. The average U.S. household’s annual income is $126,700, according to the Census Bureau. That household would need nearly 16 million years in order to reach $2 trillion, and then only if it saved every single penny.”
Stop right there and read that bolded sentence again.
The average U.S. household’s annual income is $126,700.
When I read that line, my brain came to a dead halt. If you actually go outside, walk down a normal street, or have ever worked a normal job in the United States, you know immediately that $126,700 is not what an average American makes. It doesn’t even live on the same planet as what an average American makes.
Even when you point this out to standard policy wonks or economists, their first reflex is to say: “Oh, well the New York Times used the Mean ($126,700) instead of the Median Household Income (~$80,600)!”
Hold on. Even $80,600 is a statistical mirage. Why are we looking at a “household” or “income” at all? If you want to know what an average American makes, wouldn’t you just look at what an actual everyday American worker is—retail, food service, warehouse workers, cashiers, healthcare aides, drivers, office clerks—and look at their individual wages?
If you pull actual individual W-2 paycheck data from the Social Security Administration, more than half of all individual American wage earners make around $38,000 to $42,000 a year (roughly $18 to $20 an hour), and the single most common wage bracket in the country sits in the $30,000s.
Look at the three-stage linguistic rocket (TRUST → [OBFUSCATED MEANING] → LOGIC) required to turn a $38,000-a-year American worker into the New York Times’s “$126,700 average household”:
Obfuscation Layer 1 — Swap “A Person” for “A Household”: A household is not a human being. Because nobody can afford rent or a mortgage on a single paycheck anymore, a “household” today bundles two or three exhausted people’s paychecks under one roof—spouses both working fifty hours a week, three roommates splitting an apartment, or adult children living with their parents. Under Census Bureau math, the worse the housing crisis gets and the more working people are forced to cram under a single roof to survive, the higher “Household Income” climbs on paper!
Obfuscation Layer 2 — Swap “Wages” for “Income”: “Income” doesn’t just measure paychecks from working a job; it dumps passive capital gains, stock dividends, landlord rents, and retirement withdrawals into the same pot.
Obfuscation Layer 3 — Swap “Typical” for “Arithmetic Mean ($126,700)”: In everyday English, when a normal human hears the word “average,” they trust that it means typical. The U.S. Census Bureau itself always headlines the median specifically because billionaire fortunes break the arithmetic mean. The New York Times deliberately reached past the median and grabbed the mean ($126,700)—pouring Elon Musk, Jeff Bezos, and the top 0.1% into the numerator so their stock portfolios drag “the average household” up by $46,000 a year.
Look at what that one sentence does to a normal person reading their email at 6:51 AM before clocking into work.
First, if your family pulls in $65,000 or $80,000 a year—putting you dead-center in the real American middle—reading that “the average U.S. household makes $126,700” quietly gaslights you into thinking the economy is booming for everyone else and that you are personally failing by $50,000 a year. Second, telling a working parent that it would take their family 16 million years to match a single tech IPO isn’t designed to inform them. It is scale-terror, engineered to make the reader feel microscopic and hopeless before the workday even starts.
“In That Same Period, It Lost $80.7 Billion. But It’s Growing Fast.”
Now look back at the opening paragraph right above that $126,700 sleight of hand, sitting directly underneath the photo of Anthropic’s “Code w/ Claude” podium:
“Which artificial intelligence firm will become the biggest company ever to debut on the stock market? Anthropic offered the opening bid on Monday, when it released its plan to offer shares on the Nasdaq exchange. The numbers are staggering. In the first half of this year, Anthropic earned $16.3 billion. In that same period, it lost $80.7 billion. But it’s growing fast. Its potential valuation in the initial public offering is $2 trillion.”
Read those four words in the middle out loud: “But it’s growing fast.”
A company brings in $16.3 billion in six months and loses $80.7 billion in the exact same six months. That is $5 of pure incinerator ash for every $1 of revenue—burning through $13.4 billion a month! If you walked into a local bank and said, “Good morning! Since January I earned $16,300 and lost $80,700, so I’d like you to value my bedroom at $2 million,” security would tase you in the parking lot.
When Anthropic does it on a Monday to pitch the biggest IPO in human history so early venture capital and defense insiders can tap public retirement funds for exit liquidity, the New York Times shows up on Wednesday morning and waves away an eighty-one-billion-dollar crater with four words: “But it’s growing fast.”
Why “Is It an AI Bubble?” Is Propaganda for a $3 Trillion Q-Day Grid
Right after pitching the $2 trillion IPO, Sam Sifton sets up the frame for the rest of the newsletter:
“As companies and investors wrestle with the colossal size of the A.I. economy — and as the markets they serve grow anxious about whether it’s all just a bubble, sure to burst — Evan Gorelick and I thought an index would be a useful way to examine what’s happening. Numbers (even ginormous ones!) can help us understand what’s happening in the world.”
Then they print the index itself:
Amount of money Amazon, Google, Microsoft and Meta are projected to spend on A.I. infrastructure this year: $700 billion
Amount of debt taken on by hyperscalers, chipmakers, neoclouds and other companies providing data centers and electricity for the A.I. build out: at least $3 trillion
Share of projected S&P 500 earnings gains for this year driven directly by tech companies spending heavily on A.I.: two-thirds
Share of the world’s electricity that data centers currently consume: 1.5 percent
Share of the world’s electricity generated by Germany: 1.5 percent
Look at those five numbers together, and you immediately see why the media’s favorite question—“Is it an AI bubble?”—is a distraction.
Calling this an “AI bubble” assumes that institutional investors are treating this like a normal commercial software investment, betting that $20-a-month chatbot subscriptions will magically pay back $3 trillion in debt. Nobody takes on $3 trillion in debt, spends $700 billion a year in capex, and buys the entire electrical output of Germany (1.5% of the planet’s power) just to help office workers summarize emails at a loss.
The smart money isn’t funding a software startup boom. They are funding Q-Day infrastructure—the massive, centralized compute fortresses, high-density fiber interconnects, and dedicated gigawatt power grids required for post-encryption dominance (Quantum Day, the threshold where quantum-scale compute shatters legacy encryption and whoever owns the physical compute-and-power rails owns the ledger of the modern world).
Look at what the “Boom vs. Bubble” false binary actually accomplishes:
If you believe the “Boom” story, you pour your 401(k) into a $2 trillion IPO that loses $80.7 billion every six months.
If you believe the “Bubble” story, you sit back on the couch and shrug: “Haha, dumb tech bros, it’ll pop on its own just like Pets.com in 2000.”
Except concrete data centers and power plants don’t pop when a stock price drops. Commercial “AI” is simply the public wrapper used to get two-thirds of the entire S&P 500 to subsidize and normalize building a $3 trillion physical compute-and-energy monopoly. Even if the stock market valuation “bursts” tomorrow and wipes out retail shareholders, the physical Q-Day fortresses, the nuclear and grid hookups eating Germany’s worth of electricity, and the tollbooths sitting over the economy remain permanently in the hands of the four monopolies that built them.
The Worker Punchline
(And the 10-Minute Yoga Sedative)
And what happens to the actual human beings inside that grid while the concrete is being poured? The New York Times prints the answer right in the same list, dressed up as breezy morning trivia:
Workers who reported using A.I. as part of their jobs in 2023: 30 percent
Workers who reported using A.I. as part of their jobs in 2025: 76 percent
Estimated number of American workers exposed to A.I. who may need to transition to new jobs: 11 million
Slowdown since 2023 in real wage growth for jobs highly exposed to A.I., compared with ones that are not: 6.7 percent
Number of federal laws passed to regulate the development of A.I. technology: 0
Put the two halves of the Times’s own list together:
Workplace AI adoption jumped from 30% to 76% of workers in two years, and AI spending accounts for two-thirds of all S&P 500 earnings gains.
And what did the 76% of workers who actually use the tools to generate that productivity receive? 11 million of them are lined up to be displaced, and the ones still sitting at their desks got hit with a 6.7% slowdown in real wage growth—with zero federal laws passed to regulate any of it.
Because the media calls this “private tech innovation” instead of a civil-scale infrastructure grid, three-quarters of the workforce gets locked onto the rails, their real wage growth drops by 6.7% to help subsidize the $3 trillion build-out, and Congress passes 0 laws.
Even better, if you scroll just three inches further down the exact same email—past a blurb about North Carolina residents teetering across bare iron beams because a hurricane washed away their bridges—you find this single bullet point tucked under Midterm Elections:
“Two dozen of the world’s richest people or their advisers have been meeting weekly — and raising hundreds of millions of dollars — to defeat California’s proposed ‘wealth tax,’ which will be on the ballot next week.”
Twenty-four billionaires holding weekly group-project meetings to spend hundreds of millions of dollars making sure they never pay a wealth tax, right beneath an index showing their companies taking on $3 trillion in debt, losing $80.7 billion in six months for a $2 trillion IPO, cutting AI-exposed worker wage growth by 6.7%, and passing zero laws.
How does a morning newsletter hand a working reader all of that at 6:51 AM without sparking a revolt in the kitchen?
Look at the rest of the scroll architecture in that email. Immediately after the index and the billionaire meetings, The Morning transitions straight into:
Recipe of the Day: Three-cup chicken.
The Morning Recommends: Boost your afternoon energy with a quick, 10-minute yoga routine.
Wirecutter Advice: Stop your home from smelling like cat litter. You don’t need to get rid of the cat!
Daily Games: Spelling Bee (yesterday’s pangram was payphone), Mini Crossword, Wordle, and Connections.
That is why the “By the Numbers” format is the ultimate modern propaganda engine. It doesn’t have to lie about the $80.7 billion loss, the 6.7% wage slowdown, or the 0 federal laws—it only has to lie about your neighbor’s paycheck ($126,700) so you blame yourself, frame a $3 trillion Q-Day infrastructure seizure as a stock-market guessing game, and sandwich your shrinking wages between a 16-million-year scale-terror comparison and a 10-minute afternoon yoga routine.
You read about your own replacement like it’s the local weather report, play the daily Wordle, and log into work.
This is The Rebel’s Contract. Your subscription isn’t a donation; it’s a direct investment in my liberation from wage slavery. Every dollar frees up my time to forge more tools for our shared arsenal. Fund the fight, and I’ll build the tools to win it.
can you please respond to my message in direct chat. New paid sub here, thanks
This is truly fantastic piece. Should be eye opening - and lead to some actual action by the populace. But we have about 275 million people completely ASLEEP in this country. And the others are too disjointed and balkanized to unite with any sort of purpose and plan in action.