Oaktree co-founder Howard Marks warns that AI’s growing demand for capital, with debt issuance nearing $600 billion this year, could reshape credit markets, push up borrowing costs and expose risks in the increasingly interconnected financing arrangements across the AI ecosystem.
Moneycontrol News
November 13, 2026 / 15:01 IST
Artificial Intelligence
The artificial intelligence (AI) boom is usually discussed as a race for chips, models and market share. Oaktree Capital Management co-founder Howard Marks believes the more consequential race may be for capital.
In his latest memo, Marks argues that AI is becoming large enough to influence not just equity valuations but also the cost of borrowing across the economy. Global AI-related debt issuance has reached nearly $600 billion this year This includes bonds issued by hyperscalers, project financing for data centres and borrowings by other AI-linked companies
Marks estimates that AI could require an average of $1 trillion in new capital annually in the coming years, with hundreds of billions of dollars potentially continuing to come from debt markets.
Economists have long used the term “crowding out” to describe how government borrowing absorbs capital that might otherwise fund private investment. Marks turns that concern around: rate-insensitive corporate borrowers, particularly hyperscalers, may now be competing aggressively for the world’s savings.
Hyperscalers cannot easily pause their AI infrastructure expansion simply because financing becomes more expensive. Slowing investment could mean losing ground to rivals. This may make them “takers” of capital, willing to pay higher borrowing costs to secure funding.
If demand for capital becomes large enough, it could push up the interest rates governments have to pay to borrow. In effect, companies building AI infrastructure may not just absorb capital but also influence its cost across the economy.
The scale of borrowing is already reshaping credit markets. Hyperscalers have grown from accounting for about 1% of issuance in the $15 trillion US investment-grade bond market to roughly 10% in two years
The shift matters because these borrowers are no longer relying solely on their cash reserves to fund expansion. Marks notes that hyperscalers are drawing on cash reserves and current cash flows while increasing their borrowings. Their investment-grade ratings make financing relatively easy today, but lenders must consider what happens if leverage rises, depreciation accelerates or AI revenues fail to keep pace with capital expenditure.
The risks extend beyond the largest issuers. Data-centre financing can depend on long-term tenant contracts, power availability, construction timelines and the residual value of chips. Some leases contain exit clauses linked to operational reliability, while certain borrowings are backed by GPUs whose useful lives may be shorter than the debt maturity.
Investors and lenders are already scrutinising these structures. Reuters has reported that banks and credit investors are examining service obligations, operational deadlines and the conditions under which large technology customers could reduce or terminate commitments.
The circular financing question
Marks also flags a less visible risk: circular transactions.
Chipmakers, hyperscalers, frontier AI labs and data-centre developers are increasingly investing in, lending to, buying from and financing one another. Money may flow from a chipmaker to a cloud provider, then to a data-centre developer, before returning as demand for the chipmaker’s products.
Such arrangements can accelerate deployment. However, they can also make it difficult to determine whether demand is genuinely independent or partly supported by vendors seeking to protect market share.
Marks questions whether these transactions would take place on an arm’s-length basis and what they mean for the creditworthiness of borrowers whose revenues depend on the same financial network.
This does not establish that the AI infrastructure build-out is unsound. It does, however, mean investors must look beyond headline revenue growth and ask a more fundamental question: who is ultimately paying for the computing capacity being sold?
Marks’ assessment of whether AI is a bubble is characteristically measured. He does not dismiss the technology as an illusion. Instead, he argues that AI is likely to transform companies, economies and everyday life in ways that are difficult to anticipate.
However, he distinguishes the technology’s potential from the economics of financing it. The key question is whether AI will generate enough value for users to sustain compute prices that deliver adequate returns on the trillions of dollars being invested.
Until that becomes clearer, he argues, much of the investment case rests on belief rather than demonstrable economics.
For bond investors, that belief has a price in the form of yields, spreads, collateral values and refinancing risks. For equity investors, it is reflected in valuations. For the wider economy, the consequence could be a higher cost of borrowing.
AI may not just absorb capital. It may change what capital costs.
Moneycontrol News
first published: Nov 13, 2026 03:01 pm
Discover the latest Business News, Sensex, and Nifty updates. Obtain Personal Finance insights, tax queries, and expert opinions on Moneycontrol or download the Moneycontrol App to stay updated!
Watch
Trending news

‘I’m getting married next week’: Qualcomm employee says manager confirmed layoff during wedding leave
- From call centre to CEO: HPE chief shares 3 things employees must do to get promoted
- Monkey snatches Rs 81,800 cash from man in Shimla, throws all notes down from tree: Video
- ‘Microsoft accidentally invented work-life balance’: Internet reacts to global Outlook outage
- WATCH: Virat Kohli separates Gurnoor Brar from Tim Seifert, tries to calm down pacer, who then gets warned by umpire
- Maatrubhumi trailer: Salman Khan’s powerful soldier avatar takes centre stage in war drama with glimpses of sacrifice and emotions
- From cinema-side cafe to Rs 700-crore biryani business: How Paradise went from serving chai and samosas to a popular restaurant chain
- Kerala class 10 student builds remote-controlled farming vehicle to ease farmers’ workload, wins Junior Science Award
- Microsoft down: Outlook, Teams and Copilot down for users worldwide as recovery continues; here’s what Microsoft has to say
- Outlook down in India: Users continue to report email service issues, outage stretches into morning
- Inside PV Sindhu’s Bengaluru home: A Mahabharata chess board table, monochrome gallery wall, private pool and more
- Inside Archana Puran Singh’s son’s newly renovated bungalow: Full-sized tennis table, plush cream sofa, retro decor and more
- He quit a Rs 1 crore Samsung job in South Korea to crack UPSC: How IIT Bombay graduate Kanishak Kataria secured AIR 1
- 5 meteors per hour to light up night sky tonight? Northern Taurid Meteor Shower peaks on November 11–12
- Sania Mirza shuts dating, second marriage buzz with straightforward statement: ‘There is no man in my life’
`);
}
if(res.stay_updated){
$(“.stay-updated-ajax”).html(res.stay_updated);
}
}catch(e){ console.log(e); }
});
},2000); // Reduced from 8s to 2s for better Core Web Vitals (LCP/UX)
});






