AI companies spent two years convincing venture capitalists that profitability was a relic of the pre-GPT era. Burned through billions. Built data centers the size of small nations. Promised investors that compute was the new oil and margins were for cowards.
Now they need debt financing.
Bond yields spiked. Turns out lenders want to get paid back. With interest. The same interest rates that apply to companies that make actual money also apply to companies whose business model is "trust us, AGI is six months away."
The AI infrastructure buildout continues anyway. These firms will pay whatever it costs because stopping now would require admitting that maybe they overspent on the first hundred billion. Sunk cost fallacy meets unlimited investor appetite meets Treasury yields that don't care about your pitch deck.
Some CFO is currently explaining to his CEO that their cost of capital just went up three hundred basis points. The CEO is nodding along. He has no idea what a basis point is. He knows it sounds expensive. He will approve the bond issuance anyway because the alternative is telling shareholders they're slowing down while Google and Microsoft sprint ahead.
Retail traders see this headline and think it's bullish. Higher costs mean higher barriers to entry. Higher barriers to entry mean less competition. Less competition means their AI stock picks are safe. They will learn what rising interest rates do to growth valuations the same way they learned what tariffs do to supply chains and what inflation does to discretionary spending.
They will lose money and blame the Fed.
Photo by Tyler Prahm on Unsplash
Leave a Comment