# AI capex — X 热门讨论 (2026-09-30 16:59 UTC)
## @jimstewartson (Jim Stewartson, Decelerationist 🇨🇦🇺🇦🇺🇸) · 09-30 14:59 · ♥61 ↻32 💬7 I’m not against AI. I made this chart with it.
But the current technology stack is almost purpose built to demand an absurd amount of energy.
Qui bono? Old computer technology companies, e.g. Nvidia Old energy companies, like Chris Wright’s Billionaires invested in old ideas
The trillions of dollars in capex being spent is obsolete the moment their monopoly of the technology stack is challenged by any number of real advancing fields, e.g. memristive, photonic, and neuromorphic.
The key problem is that the standard memory architecture for LLMs is brutally inefficient and compresses dimensionality through digitization. It’s just wrong, and in my view prevents any hope at creating a true learning system, which I believe is a precondition for general intelligence.
We’re watching the last gasp of the steam engine. > 引用 @jimstewartson: Phew. I really don’t know how to break it to America, but the economy is currently balancing on a Ponzi scheme that is on track to cost $10 trillion over the next five years—although it will never get that far.
These are conmen—just lying to everyone’s face. Every word. Every fake “calculation” is a fucking lie. There is no new “paradigm” of. computing. The LLM stack is the laziest, most inefficient hardware platform ever created. It’s absolutely insane how little innovation there is.
Elon Musk is on heavy drugs here. He’s just nodding out and mumbling his usual science fiction stories that literally make zero sense. There will be no GPUs in space. He just wants that trillionaire status back.
Jensen is trying to shove as many of his garbage GPUs into the pipeline before everyone figures out this whole goddamn thing is tulips and GPUs are a stupid ass way to do AI in the first place.
This entire industry is a malignancy created by greed and alternate reality.
It’s going to create the biggest crater in economic history because America is full of gullible morons.
Anyway. https://x.com/jimstewartson/status/2105311546847203730
## @dizaq1 (DIZ) · 09-30 14:39 · ♥32 ↻5 💬2 $JBL earnings, what it means for $SIVE brief overview!
Q4 beat: Q4 beat; FY2027 guided to 44.5b revenue, +24%, margin up to 6.1%, EPS +34% very very nice.
FY2027 guide is the headline: $44.5B revenue (+24%, ~$1.5B+ above consensus), core EPS $17.55 (+34%), core margin up 30bps to 6.1%
FY2025 net capex was just $322M just 1.1% of revenue
AI-related revenue: $13.6B in FY2026, up 50% from $9B
$JBL is positioning to growth in FY2027, they cite their strong and diverse customer and partner relationships.
"AI related demand remains very strong and continues to accelerate".
$JBL earnings deep dive and what it means for $SIVE coming soon.
Follow so you dont miss it I'm posting it soon. https://x.com/dizaq1/status/2105306463803724246
## @GreatMattsby (The Great Mattsby) · 09-30 15:44 · ♥34 ↻1 💬6 While everyone freaks out about rates, here is a chart of $ARKK and 10-year yield and why its not a dire situation anymore.
The old $ARKK / 10-year inverse correlation is dead. That's not a bug. It's the bull case.
For years the relationship was textbook. ARKK is a high-duration bet on distant cash flows ($TSLA, $PLTR, $COIN, genomics, robotics, etc.). When the 10-year yield fell, the discount rate collapsed and those names ripped. When yields rose, they got crushed.
You can see it cleanly on the chart: 2019 to early 2020, yields down, ARKK up. 2021 peak, then the 2022 rate shock that sent ARKK into a ~70% drawdown while the 10-year exploded higher.
Then something changed after the 2022 lows.
Both started grinding higher together. Yields moved from the mid-3s toward 5%+ and ARKK climbed off the floor toward ~90. That should not happen if the only thing that matters is the discount rate.
It can happen when the reason yields are rising changes.
2022 was an inflation/Fed-tightening shock. Higher rates were a tax on growth and a signal the Fed was trying to break demand. Growth stocks hate that.
The post-2022 move has been different. Yields have been pushed higher by:
Stronger real growth and investment demand, especially: AI capex Heavy fiscal supply A market that stopped pricing imminent recession
In that regime, higher yields are not just a higher discount rate. They are also a signal that expected future cash flows for disruptors are rising fast enough to more than offset the higher rate. The same forces lifting the term premium and real yields, productivity boom, massive private and public investment, are the forces that ARKK's holdings are built to capture.
That's why the co-move is bullish, not confusing. The market is no longer treating innovation as a "zero-rate toy." It's treating it as a growth engine that can live, and compound, in a 5% world.
If the old inverse correlation returns because yields spike on a growth scare, ARKK will get hurt again. If they keep rising together because the economy and the tech cycle are both stronger than people thought, the 2022 to 2026 pattern is the one that matters.
The chart isn't saying "rates don't matter." It's saying the growth is real enough that they don't matter as much.
That's the entire innovation thesis in one picture.
TLDR: Bullish https://x.com/GreatMattsby/status/2105322853151207539
## @M44_1RJ (Jimmy) · 09-29 16:39 · ♥30 ↻5 💬2 Elon is draining $TSLA cash flow on AI shit that has zero return so far, hence the valuation is dropping fast!
All by design!!! CapEx $32B?! Are you fucking kidding me?!
Where is the fucking SEC to investigate this guy?! Where’s the fucking BOD to fire this POS?!
Why’s everyone so fucking blind? He delays FSD (going nowhere) and Robotaxi while draining Tesla cash at the same time with negative earnings that can kill the company!! Do you think this is how Tesla should be run?! Then he started the rumor of folding Tesla in?!
Can you fucking smarten up and write Tesla about how you feel as a shareholder?!
Con Elon. https://x.com/M44_1RJ/status/2104974366316671275
## @menhguin (Minh Nhat Nguyen) · 09-30 15:28 · ♥32 ↻2 💬3 ive yet to see AI capex discourse about how AI capex is 1. very quick for positive ROI 2. generalised in utility across industries 3. the capex can build more of its own capex > 引用 @_AashishReddy: “Solow said that capital hits diminishing returns // But when capital can think, then the capital learns” https://t.co/Tqh0vhRKMQ https://x.com/menhguin/status/2105318799851614582