Maybe the investment grade primary market will eventually balk at absorbing a deluge of big-tech supply, but there’s scant evidence of apprehension just yet.
Last week, while reporting the briskest top-line growth since early 2022 driven in part by a 48% gain in cloud sales, Alphabet said it’ll spend upwards of $180 billion on AI-related capex in 2026, almost double last year’s outlays.
Notwithstanding Anthropic’s agentic breakthrough, Google’s enjoyed pole position in the AI race since the rollout of Gemini 3 in late November. The company’s unique capacity to seamlessly integrate Gemini into the world’s dominant search engine, to say nothing of Gmail, Workspace and so on, is a distinct advantage. And then there’s the fact that Gemini was trained on the company’s own chips.
So, there’s every reason to be optimistic, and it thus wasn’t surprising to see Alphabet’s $15 billion bond sale upsized to $20 billion early this week on massive demand, where that means the order book exceeded $100 billion. Alphabet’s paying less than 100bps above Treasurys for the longest tenor of the seven-part deal, a 40-year bond, which priced ~25bps inside initial discussions.
Suffice to say that deal went swimmingly. And Alphabet isn’t done. Next up: Five-part deals in the UK and Switzerland, where the company could probably set a record for corporate bond sales if it was inclined, given the low bar in those locales.
With the caveat that consensus 2026 cash flow estimates as they stand right now might be revised higher, projected hyper-scaler capex in 2026 is roughly equivalent to every spare dollar of cash, as illustrated above.
“With so many significant structural buyers of fixed-income, investors [are] engaged and pleased at the prospects for credit spread widening and being more attractively compensated after the poor risk-reward entry point of recent years,” Nomura’s Charlie McElligott said, referring to the prospects for heavy supply to result in at least a little more in the way of yield.
Speaking of structural buyers, Bloomberg noted on Tuesday that UK pension funds and insurers are a reliable source of demand for high quality borrowers looking to sell long-dated debt. Alphabet’s that: A great credit looking to fund for the long-term and, in the case of the franc sale, the longest of long- terms. In the Swiss deal, Alphabet will sell a 100-year note, the first such offering by a tech company since Motorola in 1997.
Needless to say, century bonds are a dicey proposition in any context, to say nothing of the tech sector, where today’s bleeding-edge innovators can be next decade’s also-rans — assuming they even live to see the next decade.
Of course, it’s hard to imagine Google ever becoming obsolete but… well, one bond manager summed it up best in remarks to Bloomberg for the above-linked article. “100-year bonds tend to have a habit of calling the top of a market.”
How much of the projected/hoped for AI driven economic growth is from AI vs AGI?
As an equity holder, I’d like to thank the lenders to Google.
I’m pretty confident that over the next 100 years, the C-Suite will be able to use the proceeds from these debt offerings for a combination of cap ex/buybacks to keep me sleeping 8 hours/night.
There was another very distinct bell that rang just this past Sunday.
Just a few weeks ago, I was thinking about the last big tech bubble. “AI might be a bubble,” I mused,” but it’s not as bad as the 2000 blow-everything-on-Super-Bowl-ads bubble.” Remember the 2000 Super Bowl? E-Trade had an ad where a monkey danced on a bucket to “La Cucaracha” while two men who appeared to have developmental disabilities clapped along. The tag line was, “Well we just wasted $2 million. What are you doing with your money?” That sure maked me trust their dedication to fiduciary responsibility.
Then there was the monster.com ad where a series of children declared their career ambitions to the camera. My favorite was a little boy who proudly declared, “I want to claw my way up to middle management!”
Then came this Super Bowl. OpenAI reportedly spent $85 million on their promotion. Anthropic had an ad which directly targeted OpenAI (it was pretty funny). Then there was a whole slew of ads from AI startups I’ve never heard of. They raised god-knows how much money from venture capitalists and rather than blowing it on NVidia chips, blew it on an advertisement. Coinbase rounded things out with the most mystifying ad of all. It was just a Backstreet Boys song, white lyrics on a blue screen like the monitor at a karaoke bar. My wife and I were both asking, “Who the hell is this an ad for?” Then it just says “coinbase” at the end. That’s it.
We are officially in the Lighting Money On Fire With Super Bowl Ads stage of the bubble. By 2000 standards, the top should be just a month away.
Have you ever heard of a company called EDS? Neither had anyone in 2000. If the purpose of an ad is to promote your brand, then EDS ad was an epic failure. But damned if they didn’t make one of the most memorable Super Bowl ads ever. The 2026 version didn’t even rise to the “People will be talking about this twenty-six years from now” level.
https://youtu.be/m_MaJDK3VNE?si=NW9ivhVDF097pSbR
That’s a good observation. The sheer number of AI ads this year reminded me of all the crypto ads that ran during Super Bowl LVI in 2022 (which some deemed the “Crypto Bowl”). That was just about nine-months before Bitcoin bottomed around $16,000. Maybe it is something like the Sports Illustrated cover jinx.
Great clip by the way. I remember my wife and I both laughing at that commercial when it aired.