The $175 Billion Question Nobody Wants to Ask

We just watched Alphabet promise to blow $175–$185 billion on AI infrastructure in 2026. Not ask for it. Promise. The market's response? Alphabet shares fell 0.5%. Then the whiplash: by Friday, when Amazon added its own $160 billion capex bomb to the pile, suddenly everyone decided this was bullish. Nvidia jumped 8% on a single day. Oracle jumped 9.6%. Broadcom extended a back-to-back advance.

Let me be direct: the market is lying to itself, and we're all pretending not to notice.

Here's what actually happened over the past week. Software stocks got decimated—we're talking 20% declines in ETF tracking the category, with ServiceNow and Salesforce falling close to 7% each on the heels of earnings that proved AI is already chewing through margin expectations. Bitcoin fell and touched its lowest level since November 2024. Private credit funds—KKR, Ares, TPG, Blue Owl—tanked double digits because they're drowning in exposure to software businesses that are about to get vaporized by better automation tools.

The economy threw a collective psychotic break. The December jobs report showed the U.S. added 50k jobs last month, with downward revisions removing 76k from prior months. The labor market, in plain English, is limping. Yet somehow we're supposed to believe that megacap tech titans are genuinely confident enough in the future of AI demand to commit hundreds of billions to infrastructure bets.

What a load.

Here's the game: when capex announcements are enormous enough, when the press releases land with enough gravitas, the investment community doesn't ask the hard question. It doesn't say: "Show us the revenue model that justifies this spend." It doesn't dig into unit economics. Instead, it pivots. It decides that announcing confidence is evidence of confidence. The narrative shifts from "software is broken, AI is cannibalizing margins" to "look at these big bets, the AI infrastructure play is still real." Supply-side enthusiasm becomes demand-side proof.

Alphabet embarked on a global bond spree to fund its AI ambitions. They're literally borrowing money to fund capex while their primary software businesses—YouTube advertising, cloud services—face structural headwinds nobody's really addressing. We're supposed to applaud the ambition instead of asking why they need to tap the debt markets if returns are so obvious.

Meanwhile, the Nikkei 225 just crossed 56,000 for the first time in history on the back of a Japanese election. Japan, a nation drowning in structural stagnation for two decades, is hitting all-time highs. The Dow reached 50,000. We're seeing the Russell 2000 up around 7.5% following today's surge, and international developed-market equities up around the 7% mark for the year, while the S&P 500 is barely clawing back to positive territory on a year-to-date basis. This isn't a bull market. This is rotation panic dressed up as broadening participation.

Small-cap leadership. Precious metals rebounding off a historic crash. Cryptoassets that fell 50% from peaks now bouncing because traders got bored being scared. Gold topped $5,000. A few weeks ago we were all supposed to believe gold was crashing on deflation fears. Now it's back at record levels because, apparently, we've decided inflation is still the risk.

The truth is simpler and meaner: the market has no idea what it believes. It's operating on pure momentum and rotation flows. One day software is toxic. The next day it's not—not because anything changed, but because the technical bounce needed a narrative, and capex announcements provided it. An ETF focused on software names extended a back-to-back advance to almost 7% by Friday. Oracle jumped almost 10%. The same stocks that got massacred are now rallying because, apparently, the massacre was "healthy price discovery."

Here's what matters in the next week: the January jobs report on Wednesday and CPI on Friday. If payrolls show real momentum—anything north of 150k jobs—then maybe the spending narrative gets real traction. If inflation's hotter than expected, rate cuts disappear and suddenly those capex bets become bets on higher cost of capital. If jobs are weak? We're about to find out what happens when confidence is a function of momentum, not fundamentals.

The Fed just signaled it's on pause, waiting to see if tariffs create "temporary" inflation. Kevin Warsh is nominated to take over Powell's chair. The policy backdrop is increasingly hawkish disguised as patient. And yet the market is still pricing in two cuts for 2026, more than the central bank has indicated it might deliver.

This is a market that's making bets on a narrative that requires everything to break in its favor. The capex announcements from the Magnificent Seven feel confident. But confidence and reality have different rent these days. And we're still not asking the question that matters: if AI infrastructure is such a phenomenal bet, why do we need quarterly exhortations to keep investors believing? Why does every earnings call need another "$10 billion more capex" like a dealer upping the dose?

Because the alternative—that we've already priced in a decade's worth of AI upside—is too terrifying to say out loud.