Hyperion irony: designed for invisibility, now on everyone's radar
Prasad Gollakota
It is unbecoming to say “I told you so”, but when it comes to Meta Platforms’ US$27.3bn project bond to fund its Hyperion data centre via private credit shop Blue Owl Capital (and SPV Beignet Investor) the temptation is strong. I wrote in late 2025 that Meta’s decision not to consolidate the debt on its balance sheet sits uneasily with the economic risk that investors have assumed they are taking when they bought the deal in October. And now Meta’s auditor, EY, clearly feels this same tension.
Although it has issued an unqualified opinion supporting Meta’s consolidation determination, EY has raised a bright red flag – aka a critical audit matter (p83).
CAMs are one of the less discussed consequences of Arthur Andersen’s collapse in the wake of Enron and the resulting creation of the Public Company Accounting Oversight Board under the Sarbanes-Oxley Act.
Created to protect investors and the public interest, the PCAOB oversees public company audits, sets auditing standards, inspects audit firms and enforces compliance. It shifted audit oversight from professional self-regulation to an independent watchdog.
The PCAOB’s regime also governs how auditors communicate risk, and under PCAOB standard AS 3101, auditors must identify CAMs, meaning judgment-heavy areas tied to material accounts or disclosures. A CAM is not a qualification to the audit opinion, but a flag that the audit involved a borderline call, and investors should read it that way.
That is the lens through which Meta’s consolidation determination should be viewed. In its 2025 Annual Report (p106), Meta says of Beignet Investor:
“We do not have the power to direct the activities that most significantly impact the venture's economic performance. Therefore, we ... do not consolidate the ... entity.”
Not an entirely comforting statement for bond investors hoping to have direct exposure to Meta credit.
EY’s audit report signed off this position with an unqualified opinion but flagged the consolidation matter as a CAM:
"Auditing the company’s determination … was especially challenging due to the significant judgment required in determining the activities that most significantly affect the [venture’s] economic performance … and assessing whether the company has the power to direct those activities."
So while EY supports Meta’s position, it does so with a warning.
Marking your own homework
At first glance, the discourse in Meta’s annual report can read like Meta marking its own homework. And you might reasonably ask: isn’t EY meant to make the determination here?
The answer is mundane but important. In a public company audit, management is responsible for preparing the accounts and making the underlying judgments. The auditor’s role is to test whether those judgments are supportable under the applicable standards, based on evidence and the actual contract terms. That is why the consolidation judgment call sits with Meta in the first instance, and why EY describes its work as an assessment: understanding the structure, reading the relevant agreements, and evaluating whether the conclusion is defensible.
The bigger issue, however, is not the accounting outcome. It's the noise.
Even if you accept that Meta is right in its assessment on non-consolidation under the rules, the transaction has now generated a level of scrutiny that was unnecessary.
Indeed, now that a key element of the transaction has been identified as a CAM, it is a signpost to interrogate it. And when Meta’s disclosure states that its maximum potential exposure over the life of the transaction is US$45.95bn, alongside the non-consolidation conclusion, it invites an obvious question: how can something be off-balance sheet while being so clearly on risk?
Incidentally, Moody's clearly has similar concerns. It was always notable that the Hyperion deal was only rated by S&P – and now we know why. Moody's on February 23 published an opinion, referencing the deal, that criticised “limitations” in hyperscaler disclosures, which “may not show the full picture” of their true liabilities.
All of which underlines the real irony of this deal. Meta does not appear to be constrained by leverage in a way that typically motivates aggressive accounting outcomes. So it has created reputational and analytical noise, paid above its curve on the bond financing (the bonds rallied by some 10 points when trading began), and has done so in a way that forces the market to debate definitions of “power” rather than focus on the underlying economics of the project.
In a world where the biggest tech firms are already under intense political and regulatory attention, self-inflicted complexity doesn’t appear sensible.
Not a template
There is a second-order effect of the CAM: as the transaction rests on a finely balanced judgment, it may not be a template for the future.
Can bankers, lawyers, private credit sponsors and CFOs treat the deal as a template to shift infrastructure financing off balance sheet while still securing access to huge amounts of data centre space and power for the next two decades?
Not reliably, as it relies on a gatekeeper accepting a close call. The next would-be issuer can copy the headline structure, but it cannot guarantee the same outcome if its auditor draws the judgment boundaries differently or simply has a lower tolerance for close calls. In the extreme, another auditor could require consolidation or even a modified opinion if it does not accept the sponsor’s position.
If another auditor lands differently on a similar fact pattern, it does not prove EY was wrong on Meta, but it does lengthen the shadow: it highlights how contingent the answer is on judgment, invites comparability questions, and keeps EY’s sign-off in the spotlight. Either way, now that the question mark is there, it does not go away.
Real power dynamics
It is easy to see why EY describes auditing Meta’s determination as “especially challenging”. It is a review made under significant pressure: a flagship client, a high-profile transaction and a structure that has been engineered to sit right on the boundary of what the rules permit.
“Challenging” can be read to mean more than accounting complexity; it can mean the discomfort of calling the boundary the wrong way. That is precisely why the CAM disclosure matters: it is the auditor acknowledging that this was not a routine box-ticking exercise.
The more interesting question that follows is not who exerts power over Beignet Investor, but how much influence a flagship client can exert over the people paid to review its conclusions.