Citadel Securities says chip-financing could hit $500bn by 2028

Citadel Securities told clients that companies could issue more than $500 billion of debt by 2028 to finance AI‑era chip purchases for data centres, a forecast that, if realised, would create a sizeable new segment inside investment‑grade credit markets.

The estimate, delivered in a client note and reported by Bloomberg on Aug. 3, 2026, comes from Jeff Eason, head of the investment‑grade desk at Citadel Securities, who said much of the borrowing would be short‑dated — roughly three to five years — and some could be placed privately as 144A offerings to speed execution and keep deals off public registers Bloomberg.

Why this matters: $500 billion of targeted chip financing would represent more than 5% of the Bloomberg US High‑Grade Index by 2028, alter issuer composition for index‑tracking funds and force portfolio managers to carve out a new industry‑specific debt bucket. Citadel frames the demand as driven by the short useful life of specialised AI accelerators and relentless data‑centre buildout by hyperscalers and chipmakers themselves LiveMint.

Jeff Eason's $500 billion forecast and direct quote

Jeff Eason told the firm’s clients that “chip financing could become one of the largest emerging segments in investment‑grade credit,” adding the current round of financing is “unprecedented relative to the current market size,” according to Bloomberg’s coverage of the note Bloomberg. Eason expects much of the borrowing to mature in three to five years — matching the economic life of high‑end AI chips — and for some issuance to come via private 144A placements to accommodate speed and covenants.

Citadel also projects chipmakers alone could account for more than $250 billion of issuance in 2028, a figure that would concentrate credit risk in a handful of large technology issuers and their suppliers LiveMint.

3–5 year debt, 144A placements and index effects

Shorter maturities are central to Citadel’s thesis. Investors typically prefer matching financing tenor to asset useful life; chips lose economic value faster than servers or buildings. That preference explains the predicted three‑to‑five‑year term, which also mitigates long‑term rate‑risk for issuers but increases refinancing frequency for corporate treasuries.

Private 144A deals, which sell to qualified institutional buyers, would let issuers avoid SEC registration and speed issuance — attractive if companies want to lock in funding quickly amid volatile markets. But 144A volume can reduce transparency for retail and some institutional investors, raising questions about market liquidity and secondary‑market pricing if volumes approach Citadel’s projection.

How the forecast compares with recent AI borrowing

Citadel’s note sits on a wave of prior AI and cloud borrowing. Some summaries suggest markets have already absorbed roughly $570 billion of AI‑related debt and about $60 billion of short‑term cloud‑giant borrowings since last year, numbers that, if accurate, show the market is not starting from zero Sina. Still, those aggregations vary by definition: analysts differ on what counts as “AI‑related” financing versus ordinary capex or inventory funding.

Skeptics also note that Citadel’s forecast depends on continued aggressive capex by hyperscalers and chipmakers; a cyclical slowdown, faster chip yields or longer useful life for accelerators would materially reduce demand. Regional reporting and some republished summaries even mangled the figure — a mistranslation in one outlet turned $500 billion into an implausible number — underscoring how sensitive interpretation of the note has become LiveMint.

Citadel’s claim is reported primarily through Bloomberg; other outlets have republished the note and added context but the forecast itself rests on the firm’s modelling and conversations with market participants rather than a public filing Bloomberg.

The near term to watch: whether issuers begin marketing multi‑hundred‑million 144A chip financings this quarter and how primary‑market bookbuilding treats three‑to‑five‑year tenors. If banks and investors treat the pipeline seriously, index providers and credit strategists will be forced to revise sector weights by year‑end.