This article is part of our new investment grade coverage — get in touch at subscriptions@9fin.com for more information on this product.

Investors have been clamoring for more investment grade tech bond supply, but now that companies are tapping both public and private markets to fund an estimated $1trn-$3trn in artificial intelligence spending the question has become — is it too much?

In recent weeks the spigot turned wide open. Tech borrowers have priced $85bn in public IG bond markets over the last two months and another $65bn has priced or is set to price in private IG debt, according to 9fin data.

Meta (Aa3/AA-), Oracle (Baa2/BBB/BBB) and Alphabet (Aa2/AA+) are the three biggest issuers leading the charge.

In mid-October Blue Owl Capital priced a $27bn joint venture private credit loan that will be used to fund Meta’s data center build out in Louisiana and then last week the Facebook and Instagram owner priced $30bn in publicly syndicated debt.

Oracle is doing a very similar package but in reverse. The cloud services company priced $18bn of syndicated debt in September and is on the verge of closing another $38bn debt package to be split up among banks to fund two data center buildouts in Texas and Wisconsin, according to reports.

This week, Google owner Alphabet threw its hat in the ring pricing a nearly $25bn dollar-equivalent bond package that included $17.5bn in the US and €6.5bn in Europe. That deal follows on from another $12.5bn dual-currency raise from April that was the company’s first since 2020.

Proceeds from the new deal are designated for general corporate purposes but are expected to be used in part to fund the $32bn acquisition of cybersecurity firm Wiz Inc., which Alphabet says will accelerate growing AI trends of improved cloud security and the ability to use multiple clouds.

Via 9fin data

Those deals bring public IG tech issuance in both dollars and euros up to $186bn (USD equivalent) year-to-date, which is already the largest year ever for tech related IG debt, according to 9fin data.

More than 45% of the year’s tech supply has priced in the last two months as hyperscalers tapped the markets to fund their data centers, so how much more is out there to get done?

“The sense that I'm getting is that this is just the initial foray because they want to build up the coffers and have enough cash that they can sustain some investment,” said Richard Wolff, head of US syndicate for Societe Generale. “So I would imagine this becomes a biannual type event, where some of these hyperscalers are in the market twice a year with large type financings depending on the growth in the space.” 

Some privacy please

Most market participants, however, say the bulk of capex funding needs will be done away from the public IG bond market.

Project financing, ABS deals, private credit trades and joint venture structures will provide different avenues for companies to diversify their investor base and even move some of that spending off balance sheet as Meta was able to do.

“My gut tells me that the funding of the AI expansion is going to be done in a multitude of ways, whether you're funding data centers through ABS, off balance sheet structures, in the bank market, private credit market, or whether you're doing it in the public market,” said Kyle Stegemeyer, U.S. Bank's head of US IG DCM and syndicate group.

Until just a few short weeks ago, the IG bond market was largely underrepresented and under used by these hyperscalers. But the strong execution from the likes of Oracle, Meta and Alphabet might be changing that.

Includes Oracle’s private credit deal, which has not yet priced (via 9fin data)

“Oracle is one of the weakest of any of the hyperscalers from a fundamental standpoint, and they were still able to issue a large amount of debt with minimal concessions,” one IG bond investor said. “So the public markets are wide open, and the question is what are they getting in terms of either customization or pricing that kind of entice them away from public markets?”

Some of the benefits of going private include extra call options on the debt that couldn’t be achieved in public markets and the ability to hide some secrets of the trade. Likewise, ABS deals allow companies to leverage the multi-billion dollar facilities they are building for better pricing.

All of these companies are also sitting on tens of billions of dollars in cash they can tap into to fund their investments, but sources said these recent deals show that public bond issuance is still a cheaper avenue they’ll have to utilize.

“If there's something nuanced about a transaction, or a special structural feature about a transaction, the investment grade regular-way market doesn't deal great with those nuggets,” one banker said. “But if you just need straight senior capital, I think the investment grade market's still pretty hard to beat.”

Oracle, Meta and Alphabet did not respond to requests for comment.

How much is too much?

Most market participants think the bulk of funding needs will be done away from the public IG bond market. But if — for example — one quarter of the $1trn of capex funding needs comes to the bond market, then that’s $250bn of potential supply.

Some even think $1trn could be underestimating the spending. Morgan Stanley researchers have placed the figure at closer to $3trn — and if just 15% of that came to the bond market it would mean $450bn of new supply expectations.

OpenAI alone has around $1trn worth of deals with companies including Nvidia, AMD and Broadcom. Alphabet, Amazon, Meta and Microsoft each raised their capex guidance for full year 2025 during Q3 earnings reports last week to $380bn collectively, which is close to the annual nominal GDP of South Africa.

More of that spending could hit the IG bond market given that Amazon hasn’t issued in the IG bond market since 2022 and Microsoft trades tight of US government debt since it last priced corporate bonds in 2017.

Amazon’s capex expenses are expected to climb to around $125bn this year. Microsoft already spent $72.6bn for the fiscal year ended in June 2025 and expects that to grow to around $91bn in fiscal year 2026. In fact, Microsoft has already gone above its previous guidance to spend $34.5bn in the quarter ended in September.

GDP data via World Bank, IMF and CIA World Fact Book. Oracle and Microsoft capex estimates based on guidance for the fiscal year ending June 2026, all other companies based on 2025 full-year guidance.

“If you believe the AI trade is going to persist, and you believe in the amount of investment that's needed, not only by the hyperscalers, but also with the tertiary industries, then you should expect continued supply to help fund that,” U.S. Bank’s Stegemeyer said.

This is just the start as most people think 2026 will be an even bigger funding year for tech, sources told 9fin. Analysts at Morgan Stanley said in a recent note that they expect total hyperscaler capex to grow 24% next year to nearly $550bn.

“Every one of these companies could do $75bn to $100bn in the bond market — I mean Meta just doubled its debt stack and its leverage barely went up so it's not about overall debt capacity,” one portfolio manager said. “What's the cadence of this issuance? Are we going to see a hyperscaler deal every quarter for $25bn? That's something the market would have to kind of plow through.”

Demand has been insatiable all year so all this extra supply is welcome to this point. But some are wondering what happens if yields come down, or there’s a bigger market disturbance.

All of a sudden, the hyperscalers could crowd out other borrowers who are not in the AI space and force them to pay more to access the market.

“Inflows can turn into outflows, and then you've got a problem where maybe the demand is not as great as it once was at a time when supply due to AI related financing is enormous,” said Peter Higgins, head of fixed income for Shelton Capital. “So you might have a mismatch, and it could very well crowd out other borrowers from the public markets.”

Where’s the hyper revenue?

Of course, the fear with all of this spending is that it’s fueling an AI bubble.

To this point there hasn’t been much transparency around how much revenue AI is generating or how much is needed to justify these investments.

A study from Bain & Company puts the target at $2trn of additional annual revenue globally — or $800bn when accounting for potential cost savings. Annual AI revenues stand at only $20bn currently, which would require a 100-fold increase in revenue between now and 2030, according to Morningstar.

“I do feel like at some point in time for our marketplace, they're gonna have to demonstrate a return on this investment,” the portfolio manager said. “But it doesn't have to be within the first innings of this spending.”

Even if it’s a big AI bubble, most people believe the hyperscalers will do just fine. The balance sheets are still incredibly strong. It’s the other companies surrounding this AI phenomenon that could be in bigger trouble.

“If we all decided tomorrow AI is dead, it's just not going to happen — it wouldn’t be the end of any of these companies,” the portfolio manager said. “But it would be the end of some of these smaller companies that have just kind of grown up as tertiary businesses around AI. Those are the companies that I think might have a problem down the road, if this doesn't go as swimmingly as everyone thinks it might.”