Bank of America sees bonds becoming the hedge against an AI selloff

For most of the year, AI anxiety and rising bond yields traveled as a pair. Investors fretted about the AI trade, and yields climbed right alongside the worry.

That relationship may be breaking down. Bank of America’s derivatives team now argues that if the AI bubble pops, stocks and yields could fall together. In that scenario, Treasuries go from casualty to cushion.

The OpenAI revenue scare that flipped the script

The trigger was a disappointing revenue figure from OpenAI in October 2026. The company’s annualized net revenue was estimated at $50 billion. Earlier projections had pointed to $70 billion.

A $20 billion gap tends to get noticed, and markets noticed fast. The Nasdaq dropped more than 1% in a single session.

The AI hardware and infrastructure names took the hardest hits. Nvidia slid 2.9%, AMD fell 3.9%, and Oracle dropped 5.5%.

The more interesting move came in fixed income. US Treasury yields declined at the same time equities sold off.

That is a departure from the 2026 pattern. Earlier in the year, AI-related fears had generally pushed yields higher, not lower. Bonds were part of the problem, not the shelter from it.

BofA’s read is that the market reaction tells you how investors are now pricing an AI slowdown. The bank frames it as a demand shock: a hit to growth that could compress equity prices and yields simultaneously.

Based on that view, BofA recommends that investors put on a hedge for this specific risk: an AI-driven growth shock that drags down stocks and bond yields in tandem.

Fund managers were worried about the opposite problem

The timing makes BofA’s call more notable. Just a month earlier, the professional money crowd was focused on a very different threat.

In BofA’s September fund-manager survey, 33% of respondents named rising bond yields as the top tail risk. AI bubble concerns came in second at 28%.

The OpenAI episode suggests those two fears may not behave the way investors assumed. If an AI bust pushes yields down rather than up, the two top worries on the list could partially offset each other.

The debt pile behind the AI buildout

There is a reason the bond market sits so close to the AI story. The infrastructure behind AI is expensive, and a meaningful chunk of it is being financed with borrowed money.

Hyperscalers issued $121 billion in corporate bonds in the year leading up to October 2026. Hyperscalers are the giant cloud operators building the data centers that train and run AI models.

BofA expects that borrowing to keep growing. The bank forecasts annual corporate bond issuance from hyperscalers for AI infrastructure of between $140 billion and $300 billion over the next three years.

So the AI trade is not just an equity story. It is increasingly a credit story too, with corporate America leaning on capital markets to fund the buildout.

What this means for investors

The core shift is about correlation, meaning how assets move relative to each other. A classic stock-and-bond portfolio works best when the two move in opposite directions. Earlier in 2026, AI fears had been pushing them the wrong way for that setup.

If BofA’s thesis holds, Treasuries regain some of their traditional role as a shock absorber, at least for AI-specific selloffs. An investor heavily exposed to Nvidia, AMD, or Oracle could see bond holdings rise in value just as those positions fall.

The hyperscaler borrowing wave is the other wildcard. A forecast of up to $300 billion a year in AI-related corporate bonds means a lot of new supply hitting credit markets.

For now, the takeaway from BofA’s derivatives desk is that the market has quietly rewired its reaction to bad AI news. Stocks still fall. Yields, which used to rise in sympathy with the panic, now appear willing to fall with them.