(Almost) Daily Musings

Evening all

80% of New Year’s resolutions don’t make it past January. I can attest—it’s not easy. Dry Q1 is already looking precariously fragile as I stroll past the Moseleck. The siren songs of lukewarm beer, a jukebox that devours euros like OpenAI burns cash, and a dartboard determined to prove, once and for all, that glasses or not, I should be legally blind.

With my resolutions having survived this hardest of tests, I am keen to kick the new year off with some musings!

Cheers
Phil

Today’s Takes

A(pple)I Crumble

Big Tech’s obsession with AI is the perfect trifecta: massive investment, fleeting profits, and inevitable market correction. A combination that, historically, has never not ended poorly. The lesson? Learning from past mistakes seems to be the actual innovation gap. Over the past five years, tech giants have collectively increased AI spending by 250%, with total investments projected to hit $280bn this year. The punchline: free cash flow growth has turned negative for several of these companies. It’s almost as if throwing absurd amounts of money at the “next big thing” doesn’t guarantee a home run. But hey, as the Baltimore Orioles learned when they handed $161m to Chris Davis (for the worst batting average in MLB history), throwing cash around is not always a winning bet.

Data snacks

• Projected combined capex for major tech companies in 2026: $336.5bn

• Sequoia Capital's revenue target to justify AI investments: $600bn

MSFT is diving in headfirst, pledging $80bn in fiscal 2025 for AI-enabled data centers. Over half of that pegged for the US, per their recent blog post. The official narrative: “train AI models” and “deploy cloud applications worldwide.” The unofficial reality? It’s starting to feel more like a territorial land grab, especially since MSFT now lists OpenAI—its former BFF—as a competitor in an SEC filing.

Even OpenAI, the prom queen of generative AI, is sweating. Their Pro subscription, priced at $200/month, isn’t making money. Why? Too many people are actually using it.

OpenAI, once joking that it would “build a generally intelligent system to figure out how to make money,” now seems content testing how much cash it can burn. Despite a $157bn valuation, OpenAI expects $3.7bn in revenue by the end of 2024—against $8.7bn in costs. A $5bn loss isn’t exactly ground-breaking.

But this AI gold rush is starting to feel a lot like the solar energy frenzy of the 2010s. Back then, companies overbuilt, flooding the market with solar panels. Prices collapsed, inventory piled up, and margins shrivelled. AI could face a similar fate. Despite the noise, only 6.1% of U.S. businesses were using AI as of late 2024—a modest increase from 3.7% the year prior. Meanwhile, demand for AI infrastructure has sparked an energy crisis: AI data centers could consume up to 17% of U.S. electricity by 2030. That’s not just unsustainable; it’s downright precarious for an already fragile grid.

All told, Big Tech’s AI ambitions might just be a contrarian investor’s dream. The solar and shale industries taught us that rapid expansion often ends in painful contractions. Yet, the lessons seem ignored—if not deliberately erased—from the wall. If AI fails to deliver the gargantuan returns investors are banking on, these cutting-edge data centers could become nothing more than hulking monuments to miscalculation.

It’s a thought that struck me as I travelled back down to mid-south Germany after the Christmas break in my northern hometown: what goes up must come down—sometimes quite literally.

Yield Yanking

The spread between 10-year and 2-year Treasuries is the widest since 2022, thanks to a bear steepener—where yields are rising across the board, but long-term bonds are selling off harder than short-term ones.

Translation: Investors are pricing in a new normal for rates and inflation worries persist:

1. Fed Funds at 4% Forever: The market seems oddly comfortable with the idea that 4% is no longer restrictive. It’s fine. Supposedly.

   

2. Term Premium Returns: Longer-term yields are climbing as fiscal uncertainty (read: deficits + tariffs + tax cuts) makes the long end a riskier bet.

   

3. Unlike the US, China’s 2/10 curve is flattening. The deflationary expectations seem to tell by contrast that the inflation expectations are a key contributor in the current 10Y performance. While the US steepens on inflation and procyclical fiscal policies, China’s curve is reflecting growth concerns and tariff-induced deflation. Context:

   1. For the US: Tariffs inflate. No more cheap imports = higher prices. This adds to inflationary pressures, pushing long-end yields higher.

   2. For China: Tariffs deflate. To keep exports flowing despite higher costs, Chinese manufacturers are price-cutting, attempting to dump goods into global markets.

   

(Almost) Enjoyed These Takes?

Pass it along—via Email, Twitter, or LinkedIn—to anyone whose inbox could use a little more muse!