There’s an ETF for That? AI Bonds

Debt issuance among hyperscalers and neoclouds has increased throughout the year, with a number of ETF issuers filing or launching new funds to capitalize on this dynamic.

By Tony Dong, CETF®, Lead ETF Analyst · November 6, 2026

Keep up with what matters in ETFs

Get timely ETF insights, market trends, and top ideas straight to your inbox.

Your newsletter subscriptions with us are subject to ETF Central's Privacy Policy and Terms and Conditions.

For much of the artificial intelligence (AI) boom, investors have focused on the equity side of the trade. Semiconductor manufacturers, data center operators, power infrastructure providers and other so-called picks-and-shovels and bottleneck businesses have attracted enormous amounts of capital. But the scale of borrowing among hyperscalers has accelerated considerably throughout 2026.

According to Bloomberg, Alphabet raised nearly $32 billion in debt over a 24-hour period in February, including a $20 billion U.S. dollar bond offering and additional sterling- and Swiss franc-denominated issues. The Google parent subsequently returned to the market with a $25 billion investment-grade offering in August, followed by a €6 billion sale in November¹.

Meta Platforms has been similarly active. After issuing $25 billion in bonds in April, it raised another $15 billion through a five-part U.S. dollar offering on November 3. Bloomberg reported that the latest transaction attracted as much as $58 billion in investor orders, suggesting considerable demand despite growing concerns about the economics of AI infrastructure spending².

A newer category of neoclouds has also been tapping debt markets to finance the GPUs, data centers and networking equipment needed to supply computing capacity. However, the credit profiles of these businesses can look considerably different from those of established hyperscalers.

Consider CoreWeave, which reported approximately $35.6 billion in total indebtedness as of June 30, 2026. Much of its expansion has been financed through borrowing backed by computing equipment and customer contracts, alongside unsecured bonds and other financing arrangements.

Bond investors have become increasingly concerned about the associated leverage. Bloomberg reported that CoreWeave's unsecured notes maturing in 2032 fell below 80 cents on the dollar in October, pushing their yields above 15%, a level commonly associated with distressed debt³.

All in all, according to FTSE Russell, investment-grade hyperscaler bond issuance had already reached $218.1 billion by early October 2026, more than double the $107.3 billion issued throughout 2025. By the end of September, hyperscalers accounted for 4.7% of the FTSE U.S. Broad Investment-Grade Corporate Bond Index, up from 2.7% two years earlier⁴.

Historically, sector-specific bond ETFs haven't attracted nearly the same attention as their equity counterparts. Now, some issuers are attempting to change that with thematic fixed-income ETFs focused specifically on financing the AI buildout.

As regular readers of this column will recognize, yes, there's an ETF for that. Today, we'll examine the AI-focused bond ETFs already available on the market, along with several upcoming filings worth keeping an eye on as issuers look to capitalize on this emerging segment of thematic fixed income.

Report

Global ETF Survey 2026 is Live!

With 1,790 new products launched by July 2026, choosing is harder than ever. Our Global ETF Survey 2026, built on 620 professional investors and allocators, shows how active, fixed income, options-based and thematic ETFs are reshaping fund selection.

Explore Results

What's Available on the Market Right Now?

The most prominent example so far is the Defiance AI Infrastructure Bond ETF

AIBN

-0.04%

⁵, hich debuted on October 19, 2026, as the first U.S.-listed ETF dedicated to debt issued by companies involved in AI infrastructure. AIBN follows a rules-based strategy with monthly rebalancing, dividing its portfolio into two distinct allocations.

The first 50% is reserved for bonds issued by four major hyperscalers: Amazon, Microsoft, Alphabet and Meta Platforms. Each receives a fixed 12.5% portfolio allocation, with up to 15 bonds per issuer selected and weighted by market value. This forms the portfolio's core, providing exposure to established technology companies with substantial operating cash flows and investment-grade credit profiles.

The remaining 50% is allocated to a broader collection of semiconductor and AI computing infrastructure companies. Eligible issuers include constituents of the MVIS U.S. Listed Semiconductor 25 Index, alongside companies involved in AI computing and infrastructure, such as IBM, Oracle, CoreWeave and SpaceX. This satellite allocation is market-value weighted, subject to a 5% cap per issuer.

The methodology targets at least 20 issuers and approximately 200 bonds across the portfolio, helping limit concentration outside the four hyperscaler positions. Eligible securities must be fixed-rate, U.S. dollar-denominated corporate bonds with qualifying maturities and a rating from at least one agency.

Importantly, the methodology doesn't restrict the portfolio to investment-grade debt rated BBB/Baa or higher. High-yield bonds are also eligible. That makes AIBN an interesting hybrid. Half the portfolio is anchored in the debt of four financially established hyperscalers, while the other half can reach further into the AI infrastructure supply chain, including issuers with more speculative credit profiles.

Because AIBN is so new, it doesn't yet have a published 30-day SEC yield. My expectation is that its yield could eventually fall somewhere between those of comparable investment-grade and high-yield corporate bond ETFs, depending on the actual credit mix, maturities and prevailing market yields.

The fund charges a 0.41% expense ratio, which is considerably higher than broad corporate bond index ETFs. Investors are effectively paying a thematic premium for targeted AI infrastructure credit exposure, rather than obtaining that exposure indirectly through a diversified bond portfolio.

What's Coming Up?

Defiance may have been first to the market, but other issuers are close behind. Three notable filings worth watching come from ProShares, Bitwise and BondBloxx, each taking a somewhat different approach to investing in AI-related debt.

The first is the proposed ProShares AI Hyperscaler Bond ETF⁶, which would track the Verify AI Hyperscaler Bond Index. This is arguably the most straightforward of the three strategies. The index identifies five major companies: Microsoft, Alphabet, Oracle, Amazon and Meta Platforms.

Eligible securities must be fixed-rate, U.S. dollar-denominated bonds with at least one year remaining until maturity. Convertible bonds are excluded, and qualifying securities are weighted by market value.

Compared with AIBN, ProShares' proposed fund would offer a narrower interpretation of AI-related debt, concentrating on the hyperscalers themselves rather than extending into semiconductor manufacturers, neoclouds and other infrastructure providers.

Next is the proposed Bitwise AI Bond ETF⁷, which is expected to list on NYSE Arca. Unlike the rules-based approaches from Defiance and ProShares, Bitwise plans to use active management. This gives Bitwise considerable flexibility to invest across different parts of the AI financing ecosystem, including businesses involved in AI models, computing infrastructure, data center ownership and project financing.

Its investment universe is also considerably broader, encompassing conventional corporate bonds, asset-backed securities, commercial mortgage-backed securities and debt issued through special purpose vehicles, joint ventures and other project-finance structures.

Under the proposed methodology, an issuer can qualify as an AI company if at least 50% of its revenue, profits or operating income is attributable to qualifying AI activities, or if at least 50% of its consolidated assets or research and development spending is associated with those activities. The definition also encompasses companies engaged in large-scale data center and cloud operations.

Portfolio construction would combine top-down allocation decisions with bottom-up credit research. At the market level, managers would evaluate relative valuations, bond supply, credit spreads and ratings. At the issuer level, they would examine contracted revenues, lease terms, tenant and counterparty quality, asset locations and technological obsolescence risk. They would also assess structural protections such as covenants, amortization schedules and available liquidity.

I think that last part is particularly relevant to AI infrastructure lending. A data center or GPU-backed financing arrangement can look attractive based on contracted cash flows, but the quality of those contracts, the creditworthiness of the customer and the potential depreciation of the underlying equipment can materially affect repayment risk.

Finally, dedicated fixed-income ETF issuer BondBloxx has filed for the BondBloxx AI Infrastructure Debt ETF⁸. This proposed fund would also have considerable flexibility in selecting securities. Its mandate extends beyond investment-grade corporate bonds to include high-yield debt and leveraged loans.

BondBloxx defines AI infrastructure broadly, encompassing data center owners, developers and operators, neoclouds, hyperscalers, power and energy infrastructure providers, semiconductor equipment manufacturers and enabling software companies.

One particularly interesting feature is its ability to allocate up to 20% of assets to non-U.S.-dollar-denominated debt and/or international issuers. That could provide exposure to financing activity outside the U.S. market, where companies are also investing heavily in data centers, electricity generation and other infrastructure supporting AI adoption.

Please note that this article reflects the author’s personal views and does not represent the opinions of the publication or its affiliates. It is for informational purposes only and does not constitute investment advice. It is essential to seek guidance from a registered financial professional before making any investment decisions.