AI capex · X 热门 · 2026-09-26 03:41 UTC

# AI capex — X 热门讨论 (2026-09-26 03:41 UTC)

## @_The_Prophet__ (SightBringer) · 09-26 02:05 · ♥57 ↻10 💬6 ⚡️The marginal Fed hike looks like a mistake.

The problem is that the transmission mechanism has become badly mismatched to the economy.

The biggest new source of investment demand, AI, compute, power, data centers, chips, transmission, is strategically compelled. The expected payoff is so large that a few hundred basis points of financing cost does not shut it down. The hyperscalers keep building.

Meanwhile the sectors that are exquisitely sensitive to rates get crushed first.

Housing.

Commercial real estate.

Small business.

Startups.

Leveraged companies.

Anyone refinancing.

So the Fed can keep raising rates and destroy increasingly large pieces of the ordinary economy while the very investment boom keeping aggregate demand strong continues almost untouched.

That is the fracture.

Then the second-order effects start fighting the Fed.

Higher rates raise Treasury interest payments.

Those payments become income for bondholders, money-market funds, wealthy households, and cash-rich corporations.

Higher rates make new housing and infrastructure more expensive to build.

Higher rates raise the hurdle rate for new power generation, transmission, factories, and other supply-expanding investment.

So the Fed can simultaneously weaken demand in fragile sectors, increase income flowing to capital owners, and make future supply more expensive.

That is a very different economy from the textbook model.

And the energy shock makes the mismatch worse.

If diesel, gasoline, crude, electricity, or other physical inputs are pushing prices higher, rate hikes do not manufacture energy. They mainly destroy enough unrelated demand elsewhere to offset the supply shock.

That is an extraordinarily expensive way to fight inflation.

The deeper danger is this:

AI can keep the economy looking strong long enough for the Fed to overtighten everything outside AI.

That delays the visible break.

GDP holds up.

Capex holds up.

Mega-cap earnings hold up.

The Fed interprets resilience as room to keep tightening.

But underneath the aggregate numbers, housing freezes, credit deteriorates, hiring weakens, refinancing pain compounds, and fiscal interest expense accelerates.

Then eventually the thing breaks somewhere the Fed was not trying to break.

That is the setup.

Ackman’s most important insight is that the economy is no longer responding uniformly to the price of money.

There are now two monetary sensitivities living inside one GDP number.

One side is strategically compelled to spend.

The other side is getting strangled by the cost of capital.

That means the Fed has to apply more pressure to produce the same aggregate slowdown.

More pressure means more collateral damage.

And eventually the policy becomes self-defeating because the sovereign itself starts absorbing more and more of the cost through interest expense.

So the highest-coherence path is:

AI capex stays strong.

The Fed remains tighter than the ordinary economy can comfortably bear.

Housing and credit weaken further.

The fiscal interest burden keeps rising.

Inflation falls more slowly than expected because energy and supply constraints remain alive.

The Fed stays restrictive too long.

Then the deterioration finally becomes broad enough that policy has to reverse harder than it otherwise would have.

That is when real yields roll over and the repression thesis moves from theory toward policy reality. > 引用 @BillAckman: The presumption that the Fed raising short-term rates reduces inflation is predicated on the belief that higher rates reduce demand and investment.

But what if higher rates don’t reduce demand and investment because the demand for intelligence and energy is unaffected by higher rates because winning the race for super intelligence has a near infinite ROI and the demand for compute will remain incalculable.

Why won’t higher rates at this unique moment in history therefore lead to more inflation as interest costs are embedded in everything?

And the problem is compounded as the more the Fed raises rates, the more inflation we will have and the more the Fed will need to raise rates further and so on.

But what if the old models don’t apply to the current paradigm and the Fed is wrong?

I think the Fed might have just made a mistake. Am I right or am I wrong? https://x.com/_The_Prophet__/status/2103667158605758704

## @trevornoren (Trevor Noren) · 09-25 17:00 · ♥41 ↻12 💬9 AI's concentration risk: "Top 10% of customers account for 99.5% of model-serving spend and 99% of neocloud spend, leaving the bottom 90% of firms with 0.5% and 1%...The bottom line is that adoption is broadening while the spending base is not, and AI infrastructure will keep depending on a small set of heavy spenders until the tail scales up."

This is certainly evidence of the technology's immaturity—over time the spending base will expand as more companies figure out how to effectively integrate AI to unlock operational value. However, it's also evidence that adoption challenges are far more persistent than the model builders anticipated. I quoted Sam Altman on this in my recent report on "The AI Trade" (https://t.co/wQQNniS2kj): "The economy just has so much inertia. People just keep doing the same things. They keep buying from the same company. They keep using their tools in the same way. I think that’s actually a positive in many ways. It’s going to make this big transition in front of us go smoother and slower. But I think it means we’ve all been too ambitious on timelines."

It's not just about inertia. AI is still plagued by its weaknesses, from hallucination to agentic workflows breaking down midstream. But to the inertia point, AI puts unprecedented transformational demands on enterprises. As I warned in my December report on "GenAI & Productivity" (https://t.co/F85Yr41edt):

"As much attention was paid to the headline 95% failure estimate by MIT researchers, their explanation for that failure rate was likely a more important long-term consideration in understanding when and how companies will realize productivity gains from genAI. To quote the researchers: “The dominant barrier to crossing the GenAI Divide is not integration or budget, it is organizational design.” McKinsey is delivering a similar message: “Building a business for the agentic age will require a fundamental rewiring of how the business operates, innovates, and protects sources of value creation.” Deloitte is saying much the same: “This is not about adding another tool; it’s about fundamentally rethinking how work gets done from the top down.” It's difficult to look at modern history and identify an enabling technology that demanded the depth and speed of organizational transformation being suggested for genAI today."

For all of AI's capabilities, there is no path to ~$2.5t in annual AI revenue (the estimatdd amount required to offset CAPEX) unless the vast majority of enterprises become relative "heavy spenders" on a manageable timeline. Instead, the tail is elongating slowly while evidence mounts that today's "heavy spenders" are pulling back their spending. According to Ramp data, the top 1% of spenders, the cohort that drives ~80% of OpenAI and Anthropic’s enterprise revenue, cut per-employee spend by nearly 10% in August.

Chart link: https://t.co/w8hPe1w6P4 https://x.com/trevornoren/status/2103530125325463972

## @_The_Prophet__ (SightBringer) · 09-26 02:07 · ♥33 ↻2 💬9 ⚡️This is one of the strongest pieces of evidence yet that AI has stopped being a technology cycle and become a macroeconomic regime.

If that 3.63% of GDP projection is even approximately right, AI infrastructure is large enough to alter the behavior of the entire U.S. economy.

That means the AI buildout itself can keep GDP stronger, construction hotter, electricity demand higher, commodity demand tighter, corporate borrowing elevated, and long-term interest rates higher than they otherwise would be.

Which connects directly to Ackman’s argument.

The Fed is trying to cool an economy while the largest infrastructure buildout in modern American history is accelerating underneath it.

That is why ordinary monetary transmission can start looking strange.

A mortgage borrower responds to another 25 basis points.

A small business responds.

A leveraged developer responds.

A company racing for artificial superintelligence does not care nearly as much because falling behind can mean losing the entire market.

So the Fed has to apply progressively more pressure to the rate-sensitive economy to offset investment demand coming from a strategic race that refuses to slow.

That creates the split we keep seeing:

AI infrastructure booms.

Treasury yields stay high.

Housing freezes.

Small businesses struggle.

Mega-cap investment continues.

The S&P gets more concentrated.

Power and physical bottlenecks become increasingly valuable.

And capital becomes more expensive because everyone is fighting for the same pool of savings.

This also changes how to think about the bond market.

The federal government needs enormous financing.

Hyperscalers need enormous financing.

Utilities need enormous financing.

Grid infrastructure needs enormous financing.

Data centers need enormous financing.

Reindustrialization needs enormous financing.

The AI revolution is becoming a competitor to the sovereign for capital.

That is a big reason the long end can stay stubborn even if inflation eventually moderates.

There is another layer people will miss.

Infrastructure booms usually create huge social returns while destroying a lot of private capital.

Railroads transformed civilization and bankrupted plenty of railroad investors.

Telecom created the internet while enormous amounts of telecom capital were incinerated.

AI can follow the same pattern.

The infrastructure gets built.

Society receives gigantic productivity gains.

But many companies building or financing the capacity eventually discover that returns were lower than expected because everyone overbuilt simultaneously.

So “AI capex is gigantic” does not mean every AI stock wins.

It means the underlying civilization changes. > 引用 @KobeissiLetter: We are officially witnessing the biggest wave of infrastructure investment in modern US history.

Total investment in data centers and AI infrastructure is projected to average 3.63% of US GDP per year from 2025 to 2032, the highest proportion among major infrastructure buildouts since the 1800s.

The previous largest investment, railroad infrastructure, represented 2.24% of GDP per year in 1870-1890.

This was followed by highway investment that averaged 1.13% of GDP in 1956-1973, while telecommunications and fiber infrastructure averaged 1.10% in 1996-2003.

By comparison, electrification stood at just 0.50% of GDP in 1905-1925, while canal investment accounted for 0.66% in 1836-1841.

This comes as AI and data-center infrastructure investment is projected to total ~$10.3 trillion between 2025 and 2032.

The AI buildout is the largest infrastructure investment in modern US history. https://x.com/_The_Prophet__/status/2103667854147297739