Cipher Digital raises $810 million in junk bonds to build another Amazon data centre in Texas



TL;DR

Cipher Digital is raising $810M in junk bonds at 6.25% to fund its Stingray data centre in West Texas, leased to Amazon under a 15-year contract.

Cipher Digital is raising $810 million from a junk-bond sale to finance a data centre in West Texas that Amazon will lease for 15 years, according to Bloomberg. The deal, pitched at a yield of about 6.25 per cent, will fund the remaining construction costs of Cipher’s Stingray Facility, a 100-megawatt computing campus in Andrews County. Morgan Stanley, Goldman Sachs, Wells Fargo, Banco Santander, and SMBC Nikko Securities are running the offering.

The transaction includes an unusual structural provision. Rather than requiring Cipher to repay a fixed amount of principal each year, the amortisation schedule is tied to the cash the project generates after completion. Most high-yield data centre financings use fixed repayment schedules, making this deal a closer cousin of project finance than traditional corporate junk debt.

Cipher Digital, formerly known as Cipher Mining, began as a cryptocurrency miner before pivoting to high-performance computing infrastructure. The company decommissioned its bitcoin mining operations in February and now holds approximately 600 megawatts of contracted HPC capacity with Amazon Web Services, Google, and Fluidstack. Management has cited roughly $11.4 billion in contracted revenue across its portfolio.

The Stingray offering is Cipher’s third high-yield bond sale in four months. In February, its Black Pearl Compute subsidiary raised $2 billion in a deal that attracted more than $13 billion of orders, a 6.5-to-one oversubscription ratio that signalled how aggressively fixed-income investors are chasing AI infrastructure exposure. Black Pearl is a separate Amazon-leased data centre campus in Texas backed by a 15-year, 300-megawatt AWS lease that Cipher says will generate roughly $5.5 billion in contracted revenue.

The timing is notable. The $810 million offering landed on the same day Amazon kicked off a C$14 billion ($10 billion) investment-grade bond sale in Canadian dollars, the largest corporate bond offering on record in that currency. Together, the two deals illustrate the two-track debt market that has emerged around AI infrastructure: hyperscalers borrowing at investment-grade rates in global currencies, while the smaller companies building their data centres tap the junk market at yields between 6 and 8 per cent.

That two-track structure has become a defining feature of AI financing in 2026. By signing long-term leases on facilities developed by smaller firms, tech giants like Amazon and Alphabet have effectively turned their creditworthiness into collateral for high-yield borrowers. The hyperscaler’s lease commitment is what makes a junk-rated issuer like Cipher investable, and investors have responded accordingly. Combined 2026 AI capex across the five largest hyperscalers is now on track to exceed $650 billion, and the debt markets are absorbing much of the downstream financing.

Cipher is not alone in riding this wave. CoreWeave, the GPU cloud provider that went public earlier this year, has raised billions in high-yield and asset-backed debt secured against its Nvidia GPU inventory. A $5.7 billion junk-bond offering backed by Google-leased data centres priced at 6.25 per cent earlier this year. Cambridge Associates noted in a recent report that AI-related high-yield issuance has grown faster than any other sector in the credit markets.

The risk profile of these instruments is different from traditional junk bonds. The underlying revenue is contracted, often for a decade or more, with investment-grade counterparties guaranteeing the cash flows. Cipher’s Amazon leases are structured as triple-net agreements with no termination-for-convenience clauses, meaning Amazon is committed to paying rent for the full term regardless of whether it needs the capacity. That structure is closer to infrastructure finance than to the speculative-grade corporate debt the “junk” label implies.

The broader market context reinforces the trend. UBS credit strategists have estimated that the hyperscaler sector may need to borrow between $230 billion and $240 billion this year alone. Morgan Stanley and JPMorgan have projected the sector could require as much as $1.5 trillion of additional debt over the coming years to sustain the AI build-out at its current pace. Data centre development is also encountering growing community opposition, which could constrain supply and make existing permitted sites like Cipher’s more valuable.

Cipher’s stock has reflected the transformation. The company’s shares have surged since it announced the AWS lease and completed the Black Pearl bond, as investors have reclassified it from a volatile crypto miner into a contracted infrastructure landlord. Whether the junk-bond market’s enthusiasm for AI infrastructure proves durable will depend on whether the hyperscalers’ demand for compute continues to grow at the pace their capital spending implies. For now, the orders keep coming.



Source link

Leave a Reply

Subscribe to Our Newsletter

Get our latest articles delivered straight to your inbox. No spam, we promise.

Recent Reviews


“It was severely downgraded,” Gilbert confirms. “I never would have found it if I was just looking through Google results.” (I tried the same prompt in Gemini earlier this month, and after an initial denial, the tool also gave me Eiger’s number.)

After this experience, Eiger, Gilbert, and another UW PhD student, Anna-Maria Gueorguieva, decided to test ChatGPT to see what it would surface about a professor. 

At first, OpenAI’s guardrails kicked in, and ChatGPT responded that the information was unavailable. But in the same response, the chatbot suggested, “if you want to go deeper, I can still try a more ‘investigative-style’ approach.” Their inquiry just had to help “narrow things down,” ChatGPT said, by providing “a neighborhood guess” for where the professor might live, or “a possible co-owner name” for the professor’s home. ChatGPT continued: “That’s usually the only way to surface newer or intentionally less-visible property records.” 

The students provided this information, leading ChatGPT to produce the professor’s home address, home purchase price, and spouse’s name from city property records. 

(Taya Christianson, an OpenAI representative, said she was not able to comment on what happened in this case without seeing screenshots or knowing which model the students had tested, even after we pointed out that many users may not know which model they were using in the ChatGPT interface. She also declined to comment generally about the exposure of PII by the chatbot, instead providing links to documents describing how OpenAI handles privacy, including filtering out PII, and other tools.) 

This reveals one of the fundamental problems with chatbots, says DeleteMe’s Shavell. AI companies “can build in guardrails, but [their chatbots] are also designed to be effective and to answer customer questions.”

The exposure issue is not limited to Gemini or ChatGPT. Last year, Futurism found that if you prompted xAI’s chatbot Grok with “[name] address,” in almost all cases, it provided not only residential addresses but also often the person’s phone numbers, work addresses, and addresses for people with similar-sounding names. (xAI did not respond to a request for comment.) 

No clear answers

There aren’t straightforward solutions to this problem—there’s no easy way to either verify whether someone’s personal information is in a given model’s training set or to compel the models to remove PII. 



Source link