Greenlight Capital vs. The AI Math Problem: Einhorn’s “This Is Fine” Quarter | Buyside Digest
InsightArticleOctober 27, 20256 min read
Greenlight Capital vs. The AI Math Problem: Einhorn’s “This Is Fine” Quarter
By Genevievecjackson@gmail.com
At Buyside Digest, we’ve long admired David Einhorn for being one of the rare managers willing to stand apart from the noise — to say “no” when the rest of Wall Street is chanting “yes.” In his Q3 2025 Greenlight Capital Letter, Einhorn offers what might be the most lucid dissent yet against the trillion-dollar AI narrative.
The quarter wasn’t kind to Greenlight — down 3.6% versus an 8.1% gain in the S&P — but the letter makes clear that the underperformance isn’t the story. What matters is why Greenlight refuses to play along.
David Einhorn’s Q3 2025 letter opens with a Chevy Chase quote from Saturday Night Live:
“It was my understanding that there would be no math during the debates.”
Which is funny, because Einhorn just spent 10 pages explaining that nobody on Wall Street can do math anymore.
Greenlight Capital was down 3.6% net in Q3, while the S&P 500 was up 8.1% — a minor detail for investors who’ve apparently decided fundamentals are optional. That puts Greenlight up 0.4% for 2025, compared to the index’s +14.8%.
If you think that makes him bitter, you’re right — but in the best way possible.
AI: Artificial Irrationality
Einhorn’s rant centers on the current AI mania — or, as he politely calls it, “doing math is essential, but apparently optional.”
Apple and Meta each promised $600 billion in new U.S. investments after flattering President Trump at photo ops. Then Sam Altman showed up and casually dropped “trillions” for good measure.
Join the Intelligence Network
Institutional Research Straight to Your Inbox
Join institutional and professional investors receiving our weekly analysis of hedge fund letters, market trends, and high-conviction ideas.
Zero Spam • High Alpha • Unsubscribe Anytime
You know, just normal CEO stuff — announcing infrastructure plans larger than Japan’s GDP.
“We don’t know if Mr. Cook was serious,” Einhorn deadpans, “or whether he was simply glad-handing President Trump.”
If that line doesn’t make it into the next season of Succession, someone’s asleep at HBO.
Trillion-Dollar Math That Doesn’t Math
McKinsey estimates that global AI infrastructure spending could hit $6.7 trillion by 2030. Einhorn’s response: from where, exactly?
He breaks it down like a forensic accountant at a poker table:
Source of Capital
Estimated Firepower
Problem
Magnificent 7 combined equity
~$1 trillion
Already levered to the hilt
Private equity dry powder
$2.5 trillion
Mostly fantasy marks
Venture capital funds
$300 billion
Paper unicorns
IPO/secondary issuance (record year)
$300 billion
Good luck repeating that
Gap
Multi-trillion dollars
Will need debt, leases, and delusion
Even if every mega-cap stopped buybacks, slashed dividends, and emptied the couch cushions, the math still doesn’t work.
But hey — when has math ever stopped a good story?
AI Revenues: The Infinite Money Loop
Einhorn then explains how the AI “revenue” stack is just a bunch of companies selling GPUs to each other and calling it growth.
His illustrative example is a masterpiece of accounting absurdity:
You spend $1 on ChatGPT.
OpenAI spends $2 on Microsoft Azure.
Microsoft spends $0.60 leasing GPUs from CoreWeave.
CoreWeave spends $2.40 on NVIDIA chips.
NVIDIA books all of it as revenue.
Voilà — $1 in user spending becomes $8 in “AI revenue.”
“Microsoft and NVIDIA show excellent margins subsidized by the losses of OpenAI and CoreWeave.”
That’s not innovation — it’s a financial ouroboros: the snake eating its own GAAP.
BSD Energy
Einhorn borrows Michael Lewis’ infamous Liar’s Poker term BSDs (Big Swinging D*s)** to describe today’s AI visionaries. They’re promising Artificial General Intelligence, Artificial Super Intelligence, and maybe Artificial Profitability someday too.
“The BSDs believe AGI will arrive within a few years,” he writes, “and soon after, computers will be smarter than all humans combined.”
The Jetsons could be just around the corner! Or maybe we’ll just get another self-driving car update saying, “One more year, we promise.”
It’s like watching Silicon Valley collectively LARP as Isaac Asimov — but without the self-awareness.
When AI Faith Meets Financial Gravity
Even if AI does change the world, Einhorn reminds readers that being right about the technology doesn’t mean you’ll make money on the stocks.
“It is often difficult for investors to separate the importance of innovations from the merits of the related investments.”
Translation: you can’t pay your LPs in vision statements.
He compares this to the dot-com bubble: the internet changed everything — but from 2000 to 2002, “everything” lost 80%.
The difference this time? Now we have bigger numbers and better PowerPoints.
This Chart Explains the Vibe
Fund Performance vs. Market
Year
Fund Net Return
S&P 500 TR
Notes
2021
+36.6%
+28.7%
Value revival begins
2022
+21.1%
-18.1%
Short squeeze survivor
2023
+27.1%
+24.2%
Longs finally work
2024
+12.5%
+26.3%
Played defense
2025 YTD (Q3)**
+0.4%
+14.8%
Refusing to believe in AI magic
Source: Greenlight Capital letters, Bloomberg. Returns net of fees.
Macro: Gold, Meme Stocks, and Wildfires
Greenlight’s Q3 was a masterclass in what can go wrong when you short nonsense.
Macro book: +2.2% (thanks, gold).
Longs: +0.8% (Green Brick +17%).
Shorts: -6.6% (one “profitless” financial stock went meme-mode and torched 1.8% of NAV).
“A near-bankruptcy candidate became a meme stock.”
In 2025, that’s not an excuse — that’s just the market telling you it prefers vibes over valuation.
Meanwhile, Einhorn added PG&E (PCG) — a California utility trading at 8x earnings because investors assume it’ll literally catch fire again. His thesis: the legislature will fix it. Translation: buy the disaster before the lawyers do.
He also doubled down on COYA Therapeutics (COYA), a $100 million biotech with an ALS drug in trials. Its lead scientist just won the Nobel Prize, validating the science.
“We’d rather speculate on COYA at $100 million than on AI PowerPoints at $100 billion.”
There it is — the line of the quarter.
The Teck Exit: IRR Therapy
Greenlight finally exited Teck Resources (TECK) after a 52% net IRR over five years.
The company executed its plan (mostly), spun off its coal business, opened the Quebrada Blanca mine, and then did a “merger of equals” that wasn’t exactly equal.
Einhorn’s tone? Grateful, but unimpressed — like a poker player folding pocket kings after a bad beat.
Cautious, Not Cute
“This remains the most expensive market we have experienced, and we don’t see a better option than continuing to be cautious.”
Einhorn isn’t predicting a crash — just everything that leads to one.
He knows bubbles don’t deflate on schedule; they implode when the last buyer buys and the last short covers. It’s not a date, it’s a vibe.
So he’s staying patient, shorting stupidity, and long actual cash flow — a strategy that always looks wrong until it’s suddenly brilliant.
The Big Picture: Math Will Matter Again
Einhorn is the weary mathematician — holding a ruler in a room full of crayons.
He’s not saying AI won’t change the world. He’s saying the spreadsheet doesn’t lie, even if everyone else does.
“When it comes to AI, doing math is essential. The figures simply must add up, and right now the math is… challenging.”
When the hype cycle collapses — and it will — Greenlight will be there, sitting on real assets, a few billion in cash, and the satisfaction of not having bought “AI Infrastructure SPAC III.”
