Software Buyouts Hit AI Wall: Thoma Bravo Pays Up on $5B Debt

What happened

Thoma Bravo has just finished a bruising renegotiation of roughly $5 billion in debt at Proofpoint, the cybersecurity company it took private for $12.3 billion in 2021. Rather than a fresh buyout, this is an amend-and-extend refinancing that pushes the debt's maturity out to 2030 — but lenders made the private equity firm pay dearly for the extra runway 111. Goldman Sachs arranged the deal, which priced with a yield near 9.3%, adding roughly $60 million to Proofpoint's annual interest bill 678. To get lenders over the line, Thoma Bravo agreed to around 40 changes to the loan documents — up from roughly a dozen it originally proposed — including an "omni blocker" preventing asset transfers or new senior debt that could dilute existing creditors' claims, restrictions on private debt buybacks, and mandatory quarterly management calls 8919. Notably, the debt stays covenant-lite and Thoma Bravo avoided injecting new equity, so the firm retained some leverage of its own in the negotiation 8.

The irony, widely noted, is that Proofpoint's underlying credit had been improving. S&P Global Ratings had upgraded the company earlier in the summer, citing revenue growth and declining leverage 67. The tougher terms weren't a response to Proofpoint's specific performance — they reflect a market-wide repricing of software credit built on fears that generative AI tools could erode the durability of subscription revenue that made these businesses so attractive to leveraged buyers in the first place 10.

The bigger picture: a maturity wall meets an AI reckoning

The Financial Times frames the episode against a $40 billion wave of private-equity-backed software debt maturing in 2028, tied specifically to Covid-era buyouts 111. Bloomberg's own market-wide analysis puts a much larger number on the table: more than $150 billion of software-company debt — spanning leveraged loans, junk bonds, and loans held by business development companies — coming due between now and the end of 2029 10. These are not the same measurement. The FT's figure is narrower, tied to private-equity buyout debt hitting a particular refinancing year; Bloomberg's spans a wider set of instruments and lenders. Both point to the same underlying dynamic: a mountain of debt raised when money was cheap and software was treated as a bulletproof asset class now has to be refinanced in a market that no longer assumes that.

Thoma Bravo is especially exposed. Reporting puts about $9 billion of its portfolio-company debt maturing by the end of 2028, including more than $2 billion at another portfolio company, Sophos, due in March 2027 818. Sophos has already tried and failed to get private credit investors to refinance its $2.1 billion term loan, even at elevated yields, and has turned back to the syndicated loan market — with Thoma Bravo signaling it won't put in fresh equity there either 18.

Medallia: the cautionary tale sitting next to Proofpoint

Any reading of the Proofpoint refinancing sits alongside a far starker outcome at Medallia, a customer-experience software company Thoma Bravo bought for $6.4 billion in 2021. Rather than refinancing, Thoma Bravo simply handed the company to its lenders — Blackstone, Apollo and KKR among them — wiping out roughly $5.1 billion in sponsor equity, according to Reuters 16. The lender consortium injected about $150 million in new capital and cut Medallia's debt load, explicitly to fund AI-driven product investment 17. Coverage of the Medallia restructuring frames it as one of the largest private-credit workouts on record, and several outlets suggest it is now shaping how lenders approach Thoma Bravo elsewhere — creditors negotiating over Proofpoint had a recent precedent showing the firm might walk away from an overleveraged asset rather than rescue it with new money, which helps explain why Proofpoint's lenders pushed so hard for protective covenants this time 68.

How the buyout model built this exposure

Software became one of private equity's favorite hunting grounds precisely because subscription revenue looked reliable enough to support heavy borrowing, while margins were high enough to make interest payments manageable 10. In a typical leveraged buyout, the sponsor puts up only part of the purchase price and loads the rest onto the target company's balance sheet — a structure that amplifies returns if the business grows, but leaves the company obligated to service that debt regardless of how it performs 10. That was the logic behind Thoma Bravo's original Proofpoint deal, done at $176 a share, a 34% premium, after a competitive process detailed in Proofpoint's merger filings that saw Thoma Bravo raise its offer several times over roughly seven months before reaching a final price 12. At the time, CNBC called it the largest private-equity cloud deal ever, eclipsing Thoma Bravo's own $10.2 billion purchase of RealPage days earlier 13. Proofpoint kept expanding under Thoma Bravo's ownership, including a roughly $1.6 billion acquisition of Germany's Hornetsecurity that pushed its debt load to about $4.67 billion 14.

That expansion strategy — buy, add on, grow into a future exit — is exactly what's now colliding with a stalled exit market. PitchBook data show US software platform buyout value cratering to just $16.24 billion in the first five months of 2026, a pace that would put the full year at roughly a quarter of 2025's $156 billion record 15. Sponsors are shifting toward smaller add-ons and growth-equity stakes rather than large platform takeouts, and even big deals that do close are getting done with shared equity checks and minority co-investors, as with Hg's $6.4 billion take-private of Onestream 15.

Where the reporting agrees

Across the Financial Times, Bloomberg, Private Equity Wire, Reuters and PitchBook, there is no real dispute about the core mechanics: lenders have gained the upper hand over software-focused private equity sponsors, AI-driven uncertainty about the durability of subscription software is the proximate cause, and Thoma Bravo — as one of the sector's largest players — is bearing the brunt of it through Proofpoint, Sophos and Medallia simultaneously 168101618. All outlets covering Proofpoint agree on the roughly $5 billion size of the refinancing, the extension to 2030, Goldman Sachs' role, and the broad concession package lenders extracted 678919. There's also consistent agreement that software exits — sales, IPOs, dividend recaps — have all become harder, forcing sponsors toward extend-and-wait strategies rather than clean exits 11015.

Where it doesn't

The clearest divergence is quantitative: the FT's $40 billion maturity-wall figure for 2028 versus Bloomberg's more than $150 billion figure spanning 2026 through 2029 11110. These aren't necessarily contradictory — they measure different populations of debt over different windows — but neither source reconciles the two figures, and a reader could easily mistake them for competing estimates of the same exposure. Coverage also diverges on emphasis rather than fact: the FT frames Proofpoint chiefly as evidence of a sector-wide reckoning arriving on schedule, while Private Equity Wire stresses the specific irony that Proofpoint's credit metrics were improving even as its financing terms worsened 16. Bloomberg's reporting leans hardest into lender mechanics — the omni blocker, the ban on private debt buybacks, the nine days of negotiation that took proposed amendments from about a dozen to roughly 40 — details that appear in Bloomberg-sourced pieces but are treated more briefly elsewhere 8919. On Medallia, Reuters is the source for the precise $5.1 billion equity-wipeout figure, attributed to unnamed people familiar with the matter rather than confirmed by the companies themselves 16.

Taken together, the evidence best supports reading Proofpoint not as an isolated hard-luck case but as the leading edge of a pattern that Sophos looks set to repeat and that Medallia already illustrates in its most severe form. The consistency of the mechanism — lenders extracting pricing and control concessions specifically because of AI-linked doubt about software's staying power, regardless of an individual company's actual credit trajectory — is corroborated across every outlet covering the story, which is a stronger signal than any single figure in dispute.