Pinned Tweet
What's described in this thread is an incredibly uneconomical way to automate processes.
What is referred to as "multi agent systems and semantic layers" is the software doing 100x the work to get a marginally more reliable output, which introduces very un-software like variable costs to achieve the automation (on top of the base models already having extraordinary variable costs vs typical (deterministic) software).
This is primarily why token demand has exploded. Not from increased end-user demand, but from the systems requiring all of these extra layers and iterative processes just to disguise the probabilistic output in a deterministic-looking costume.
And hallucinations never go away, they just get buried under all the filters. There will always be edge cases, so even in domain specific silos humans will still be in the loop and you will never achieve actual automation.
On the cost side... using neural nets to automate is like Microsoft having to spend money every time a user enters data into a cell in excel. There are enormous variable costs to service these agents, and they're going much higher because they need to get much better to actually be useful.
Right now, venture capital and recycled megacap tech cash flows are allowing these products to be developed and delivered at a massive loss. And still, roughly 97% of ChatGPT and Copilot users alike are willing to pay exactly zero.
Heavily subsidized, and users still have little interest. The products are very very far away from where they need to be to change this. It may as well be considered science fiction. And in the event they do actually get there one day, they'll be copied by everyone else.
This is all a galactic misallocation of capital and brain power. There is no there there; another very long AI winter is coming.
Obviously.
Why on earth would they risk bringing him back when he'll be lucky to stay out of prison if it's found that he orchestrated or even just knowingly participated in the smuggling of banned chips into China?
More likely that more execs get the axe. Will be shocked if they don't pull guidance tomorrow.
BREAKING: Nvidia’s board, $NVDA, is not considering bringing Jensen Huang back as CEO, CNBC reports.
Nvidia’s stock falls to a new low of the day on the news, now down -34% from its October 6th high.
1
2
51
Pitfall Harry retweeted
"OpenAI’s model was obviously not trained by solving other Millennium Prize problems ... it generalized all the way to solving a problem no human had ever cracked."
I paused reading until he provide evidence for the above statement.
What did OpenAI train the model on?
FAA Chief Demonstrates How New AI System Reroutes Flights to Save Time youtu.be/2m-J4zYgZbg?si=hC9O… via @YouTube
1
1
425
I'm looking forward to this AI inspired mania ending, the S&P down to 5k, and $AAPL trading over $600 at 20x sales, lol.
No meaningful net selling at all. Everyone is in the pool, just swimming around to different sections.
Apple $AAPL soaring to new all-time highs and now the most valuable company in the world by more than a trillion dollars 🔥
3
102
None of this is a surprise except the 90%, which is almost certainly higher as the remaining 10% are like dominated by AI-ecosystem companies themselves who are deeply incentivized to exaggerate AI's impact.
The job cuts that cite AI directly are BS.
The AI layoff gamble is backfiring.
Companies that rushed to replace workers with AI are now cleaning up the mess, per The Guardian.
Meta scaled back plans to cut staff and use more AI agents after employees spent growing time cleaning up the agents’ technical and security errors. It’s now rehiring managers.
Microsoft and McKinsey are limiting employee AI use as costs pile up.
Walmart workers say AI scheduling and task tools set impossible standards and generate false alerts.
The payoff isn’t showing up: 90% of executives say AI hasn’t yet boosted productivity at their companies, per an Atlanta Fed working paper.
Yet more than 112,000 US job cuts this year have cited AI, per Challenger, Gray & Christmas.
71
$ANT and OpenAI perpetuals breaking down. Don't think today's bounce holds.
55
Can't believe I forgot to include elections on this list. Market clearly doesn't care about elections. Trump is a big government lefty, and now he has a Congress to match, despite their many other differences.
According to my reading of the tape...
Things that don't matter:
-Iran War/oil prices
-Fed hiking rates
-Treasury yields spiking
-equity IPOs/secondaries increasing supply
-junk spreads increasing
-tech CDS increasing
Thing that matters:
-euphoric optimism WRT AI
1
71
I wrote this a year ago. It only got worse as Trump was all over the place talking about wanting house prices higher, not lower. Later, he tried to pitch the affordability crisis as something the democrats made up.
Even without the idiotic Iran war, last night's results were inevitable.
The market doesn't care. These are short-term swings taking out stops. The AI story is what matters.
Replying to @hkuppy
"inflation", lol. People want lower PRICES. It doesn't matter if it's not rational, it's what they obviously want
And these guys are out pumping stocks & housing NON STOP. They had one chance, early, to blame recession on previous admin. It's lost
Midterms will be a slaughter.
69
$ANT perp taking it on the chin. This topped above $2.35 trillion market cap on Sep 9th...
Time to take out some market insurance...
80
Here is what these commitments look like as a multiple of free cash flow for each company (2022-25 average) ...
Alphabet - 12.1x
Meta - 18x
Microsoft - 8x
Oracle - 48.6x (!!!)
Amazon - 11.8x
Hence, maintaining a short $ORCL position as one of my only negative bets on the market right now.
The others have much more profitable underlying businesses to fall back on when this AI arms race ends in disaster.
PS: Watching the $ANT perpetuals closely as a proxy for AI euphoria breaking down. Looking a little shaky lately.
The five AI hyperscalers have almost $3 trillion in commitments that don’t show up on their balance sheets.
Whether these massive bets pay off is an open question.
1
2
258
Lol, they can "target" whatever they like. The only thing they control is what they spend to get there.
This move does show how many traders are playing this from the short side though. Very crowded and popular short, for good reason.
$ORCL - ORACLE TARGETS $225 BILLION IN REVENUE BY 2030
Oracle forecasts at least $225 billion in revenue by fiscal 2030, implying over 35% annualized growth.
The company targets 40% operating margins, $21 EPS, and $30 billion in free cash flow by fiscal 2029.
Oracle expects AI infrastructure spending to peak in 2027–2028 before transitioning to self-funded growth.
1
86
Common sense.
Coding is something like 1% of the labor force and the only enterprise use-case with even moderate penetration. And in almost all cases the coding tools are used as assistants, not replacements.
Faster AI adoption isn't leading to higher unemployment, per @RenMacLLC .
"If AI were costing jobs, the states where adoption rose most since November should be seeing jobless rates climb. They aren't. Across 46 states and DC, the correlation between the rise in AI use and the change in unemployment is -0.10, essentially nothing. For now, there's no sign that AI is pushing up unemployment where it's spreading fastest."
75
According to my reading of the tape...
Things that don't matter:
-Iran War/oil prices
-Fed hiking rates
-Treasury yields spiking
-equity IPOs/secondaries increasing supply
-junk spreads increasing
-tech CDS increasing
Thing that matters:
-euphoric optimism WRT AI
Replying to @Mr_Derivatives
This market wants very badly to go higher, still. Just speaking technically/tape reading. It trades like the whole world is short.
Getting tired of seeing charts like this all over FinX lately, promoting the idea that AI has had some profound impact on productivity. Utter nonsense...
1- the chart is a revenue chart. Employee counts are essentially unchanged at these companies over the past four years (after massive hiring during/post lockdowns, which suppressed the ratio heading into the spike). The denominator is not changing meaningfully so it's just a chart of revenue.
2- the chart spikes up 3+ years before enterprise AI is even widely available, so it's nonsense from the start to attribute the spike to AI productivity.
3- this is a chart of inflation, and the benefits of having monopoly pricing power during/after a period of deranged money printing (direct fiscal + Fed + ongoing large pro-cyclical deficits). Amazon's large retail business suppresses the effect there.
4- the current capital spending cycle is gargantuan and that money cycles through the ecosystem contributing to revenue spikes for those involved, especially those selling cloud services into said ecosystem.
Enough already. Even if this were true, coding is 1% of the labor force. Any extrapolation to the wider economy (which almost always accompanies such posts) is pure speculation if not wishcasting...
1
2
52
Microsoft buyouts explained...
This number is going MUCH higher in coming years.
BREAKING: Microsoft, $MSFT, is offering buyouts to 60,000 workers as part of the company’s efforts to cut costs amid AI adoption, per Bloomberg.
Personnel on select teams in the US, Canada, India, and the UK at a vice president level and below will be eligible.
This reflects ~27% of Microsoft's total workforce.
59
Solid thread.
"And history suggests that the longer these booms continue, the more painful their eventual collapse can be."
Correct. We are already in trouble, but the trouble is going to get much bigger by the time the market forces denouement...
Historically, falling real consumer incomes have often been a warning sign of a major recession.
Yet today, the economy remains resilient despite -ve real income growth.
Is this time truly different, or is history about to repeat itself?
A thread 🧵
1
81
Is this "transition" in the room with us right now?
AI is supposedly good at software, yet unemployment hasn't budged in that field. Apparently, the bots need to be watched pretty closely by humans. And, this is like 1% of the workforce anyway. Where is the 9% coming from?
GOLDMAN SACHS: AI COULD TEMPORARILY DISPLACE 9% OF WORKERS
Goldman Sachs estimates that a 15% productivity boost from AI could temporarily displace around 9% of workers during the transition.
Historical research shows every 1% productivity gain from technology raises unemployment by 0.3–0.4 percentage points, with effects typically fading within two years.
Occupations heavily disrupted by technology have historically lost 25% of jobs within five years.
Goldman warns AI-related layoffs could accelerate during economic downturns, amplifying labor-market weakness.
1
89
Impressive circle of life we've got going in the AI ecosystem...
⚠️ RESEARCH: 1% of companies are paying for 90% of all AI spending.
And 72 cents of every one of those dollars comes from a company that has never made a profit.
That’s @chamath breakdown. Top 1% of firms, 90% of the spend. Inside that 90%, 80% is venture money. Only 18 cents on the dollar comes from a business that actually earns what it spends.
Here’s the loop:
VCs raise a fund. The fund hands cash to a startup. The startup hands it to Nvidia, the clouds, and the labs. The labs report record revenue. The record revenue justifies the next fund.
The money goes in a circle. The profits never enter it.
So the biggest buyer of compute on earth is a balance sheet that only works if the next raise closes.
Nvidia’s real customer is the fundraising cycle.
Nobody checks whether the end customer is making money, because the end customer is the end customer’s investor.
This is the first tech boom in history where the customers don’t have revenue. When the next fund doesn’t close, that’s your AI winter.
2
79
























