Pinned Tweet
FogFisters: A Lighthouse Story, Crotch Valley Sunday, Norwegian Knockers, Bodega Bianca and the Tonsils of Edith, and My German Dexterity
171
11
11,003
What cross eyed retard would use Kalshi to buy stocks and move from $HOOD
Zero
1
94
The AI Bubble: Too Much Infrastructure, Too Much Debt, Not Enough Revenue
November 15, 2026
The artificial intelligence revolution may be real. The economics financing it may not be sustainable.
1. An unprecedented spending frenzy. S&P Global projects that six major hyperscalers will spend more than $1.3 trillion in 2027 on infrastructure. Much of this investment depends on AI demand growing fast enough to justify enormous upfront costs. That is a forecast, not guaranteed revenue.
2. Debt is exploding. The Federal Reserve's November 2026 Financial Stability Report estimates that advanced-computing debt issuance surpassed $680 billion in 2026 through October, versus $130 billion in all of 2023. Outstanding public debt exceeds $1.1 trillion, alongside trillions in additional lease commitments and contractual obligations.
3. Cash flow is deteriorating. Amazon reported negative $42.4 billion in trailing annual free cash flow, primarily reflecting accelerated AI investment, despite strong AWS growth. The industry is increasingly spending today's cash against tomorrow's expected earnings.
4. Circular financing is masking risk. Chipmakers invest in AI developers, developers buy their chips and computing capacity, and infrastructure providers borrow against those contracts. The Bank of England and Federal Reserve warn that these interlocking arrangements amplify correlated financial exposure.
5. Overbuilding could destroy returns. Companies are racing to secure data centers, electricity and GPUs before knowing long-term profitable demand. Newer chips can rapidly undermine older equipment's economics. Even if AI adoption explodes, excess computing capacity could compress prices, utilization and returns.
The central danger: This is not necessarily a bubble in AI technology. It may be a bubble in the capital allocated to building it.
Today's capacity shortages show that overbuilding is not yet proven. But if revenues disappoint, borrowing costs rise or computing prices collapse, leveraged operators could face refinancing stress, asset write-downs and forced consolidation.
AI can change the world and still become one of history's greatest capital misallocations. The internet did exactly that during the dot-com era.
36
One thing $BULL has proven exceptionally well is its ability to continuously bleed even when the rest of the market shows strength. Look at $HOOD today.
This name hasn’t just destroyed shareholder principal, it’s also inflicted massive opportunity cost by making investors miss the real AI trade.
2
482
Never forget who we are as $DUOL baggies
131
I will say this, the fact that $DUOL is staying above $100 a share after those earnings is such an epic bottom signal for the stock.
Investors know what the back half of 2026 may hold in store for this stock. There could easily be an aggressive move higher if DAU and MAU re-accelerate, while the downside remains extremely limited after already being down over 80% from all-time highs.
Arguably one of the most asymmetric bets in the market right now at its current price point. Don’t say I don’t warn you.
93
$DUOL The “would never pay or even play Duolingo when I can just use ChatGPT” is a lazy take
ChatGPT can teach you a language
So can YouTube
The question is will you actually do it every day
Duolingo isn’t selling lessons
It’s selling consistency
streaks
habit loops
progression
daily engagement
that’s the product
ChatGPT gives you infinite flexibility
which sounds great until you realize most people don’t know what to do next
what do I study
am I improving
am I doing it right
Duolingo removes all that friction and keeps you moving forward
also the AI argument is backwards
Duolingo is already using the same models
they just wrapped it in structure retention and UX that people actually stick with
this isn’t AI vs Duolingo
it’s discipline vs no discipline
and most people lose that battle
therein lies the bull case
4
232
Bears who keep insisting Duolingo is getting steamrolled by LLMs and chatbots are peddling a thesis that was never serious and has now been disproven by three straight years of data—including the Q1 2026 numbers reported today.
Revenue hit $292 million, up 27% year-over-year and ahead of consensus. Earnings per share came in at $0.89, beating estimates of $0.79. Daily active users reached 56.5 million, up 21%. Paid subscribers climbed to 12.5 million, also up 21%. Bookings grew 14% to $308.5 million. Yes, DAU growth has moderated from prior peaks and bookings growth has slowed to 14%: a reflection of the law of large numbers as the base scales into the tens of millions; but these are still massive absolute gains: millions of net new daily users and hundreds of millions in incremental bookings. These are not the metrics of a product being displaced. They are the metrics of a business that continues to add users and convert them into payers at scale.
The bear case was always built on a lazy category error. LLMs and chatbots deliver information on demand. Duolingo delivers compulsive daily habit formation through gamified streaks, spaced repetition, and behavioral design that no one-off query engine can replicate. People don’t “use ChatGPT to learn Spanish” and then stick with it for 200 straight days. They do that on Duolingo because the product is engineered to make quitting feel worse than continuing. That moat is product psychology, not raw compute or training data. The bears never understood the difference.
Even more damning for the short thesis: Duolingo is not ignoring AI; it is absorbing the useful parts on its own terms. The company has rolled out AI-powered video calls and conversation features that doubled spoken-word practice in the quarter. Users aren’t fleeing to free chatbots; they are paying Duolingo more to get structured, habit-forming practice that actually works. The numbers prove it.
The stock is down sharply from last year’s peak. Fine. Valuation has reset. Guidance was lighter than some hoped, and the market is repricing growth. But none of that validates the core bear narrative that the entire business model is toast because “AI exists.” Fundamentals did not collapse. User growth did not stall—even as the percentage growth rate moderated—and monetization did not evaporate. The only thing that failed was the 2023 short memo that assumed technology alone would erase a decade of product-market fit.
Bears who are still typing “AI disruption incoming” after today’s print are not analysts. They are coping. The data has been screaming the opposite for years. The P&L, the user metrics, and the retention curves have all been clear. The only question left is how much longer these shorts can ignore reality before the position itself becomes the punchline.
$DUOL $PLTR $HOOD
















