Teardown: Clearwater Analytics and the Enfusion Data Centers

Teardown: Clearwater Analytics and the Enfusion Data Centers

This analysis uses only public SEC filings: Clearwater Analytics' annual reports through fiscal 2025, its quarterly report for the period ended March 31, 2026, the definitive merger proxy filed April 8, 2026, the Form 8-K filed at closing, and Enfusion's annual report for fiscal 2024. I have no affiliation with Clearwater Analytics, Enfusion or any member of the buyer consortium, and no non-public information about the companies or the transaction. Where I estimate, I say so.

The two earlier teardowns each had one cloud problem. New Relic was moving out of its own data centers. Ping Identity and ForgeRock were two SaaS platforms on two different public clouds. Clearwater Analytics, taken private in June 2026, had a version of both at once, and its buyers underwrote a plan in which the fix shows up as margin expansion and nowhere as cost.

The deal

On December 20, 2025, Clearwater Analytics agreed to be acquired at $24.55 per share in cash by a consortium that included funds managed by Permira, Warburg Pincus, Francisco Partners and Temasek. A go-shop period ended on January 23, 2026, and the merger closed on June 25, 2026. According to the closing 8-K, cash paid to equityholders was about $7.4 billion, funded by about $5.7 billion of equity from the consortium and about $2.7 billion of debt.

This was not an arm's-length auction among strangers. Clearwater's board included directors affiliated with Welsh Carson, Warburg Pincus and Permira, who did not vote on the merger agreement, and a special committee ran the process with PJT Partners and J.P. Morgan as financial advisors.

Eight months before signing, Clearwater had closed its largest acquisition. It bought Enfusion, a front-office platform for investment managers and hedge funds, on April 21, 2025, for total consideration of about $1.4 billion in cash and stock. It bought Beacon Platform nine days later.

What the filings show

Both companies report on calendar years. Dollar figures are in millions.

Clearwater Analytics

Enfusion, before the acquisition

Capex is cash spent on property and equipment. Purchase obligations are future minimum payments at year end, which Clearwater describes as arrangements for hosting infrastructure services, cloud services and SaaS accounting solutions. Clearwater's 2025 figures include Enfusion from April 21.

Finding 1: One platform on AWS, one in two data centers

Clearwater's 10-K says it hosts substantially all of its cloud service in data centers operated by Amazon Web Services. Enfusion did not use a public cloud. Its last annual report says it served clients from two co-located data center facilities, in Chicago and Secaucus, New Jersey, on servers it controlled and infrastructure it custom-built. It leased data center space and dark fiber lines, and its hosting costs rose $0.5 million in 2024 as it expanded those data centers.

That is not a criticism of Enfusion. It sold a multi-tenant SaaS product, and it described its competitors as "cloud-migrated rather than cloud-native." Multi-tenant software architecture and public cloud hosting are separate choices. Chicago and Secaucus are both major hubs for financial market infrastructure, and a front-office trading and order management product may have good reasons, such as latency or market data connectivity, to sit there. The filings do not say why Enfusion chose its setup, so that is my inference.

But the combined company now runs two infrastructure models. And Clearwater's 2025 10-K states where it intends to go: Enfusion's capabilities "will be increasingly integrated" with Clearwater's to enable "a unified, cloud-native platform." Getting from two data centers to that platform is a migration, and migrations are the most expensive event in a cloud cost history. The New Relic teardown showed what one did to gross margin.

The diligence point: when an acquisition brings a second infrastructure model, the integration plan needs its own cost line. Ask whether the target workload moves to the acquirer's cloud, stays where it is, or is rebuilt. Then ask what each path costs, how long both environments run at once, and what latency or data constraints come with the workload.

Finding 2: The 2025 margin drop was mostly acquisition accounting

Clearwater's GAAP gross margin fell from 72.8% in 2024 to 67.3% in 2025, and to 65.8% in the first quarter of 2026, the first full quarter compared against a year-earlier quarter with no Enfusion in it. It would be easy to read that as a cloud problem. It mostly is not.

The 10-K breaks the $116.2 million increase in cost of revenue into parts. Depreciation and amortization, mostly from the intangible assets acquired with Enfusion and Beacon, accounted for $53.8 million. Payroll, largely from acquired headcount, accounted for $39.6 million. Data costs were $12.5 million. Technology costs were only $4.6 million.

Part of the decline is also mix. Before the deal, Enfusion's gross margin ran about five points below Clearwater's. On their 2024 standalone figures, the two companies together would have had a gross margin of about 71.2%, against Clearwater's 72.8% alone.

The diligence point: a margin decline after an acquisition is usually accounting and mix before it is operations. Separate the three before drawing conclusions, because only the operating part is something a buyer can fix.

Finding 3: The cloud bill is spread across four lines

Clearwater does not report hosting as a line item, and cost of revenue is not where most of its technology spend sits. Technology costs rose in every expense line in 2025: $4.6 million in cost of revenue, $8.4 million in research and development, $1.2 million in sales and marketing and $3.6 million in general and administrative, for about $17.7 million in total, compared with about $4.2 million the year before. The 10-K attributes the R&D increase to higher use of third-party cloud computing. Technology costs include other IT spending as well, so not all of this is cloud.

At the same time, purchase obligations rose from $12.8 million to $96.1 million. The schedule is front-loaded: $35.4 million due in 2026, $34.7 million in 2027, $24.7 million in 2028, and almost nothing after that. Those obligations include hosting and cloud services along with accounting software, and the filings do not split them out.

The diligence point: if the income statement does not show the cloud bill, it cannot be benchmarked from the income statement. Get the bill itself, every provider, every account and every data center contract, and map it to production, development and everything else. In an acquisitive company, check which contracts came with each acquisition and when each one ends.

Finding 4: The plan counts the efficiencies, not the cost

The proxy's financial forecasts take revenue from about $730 million in 2025 to about $1.8 billion in 2030, and gross margin from about 78.2% to about 80.7%, roughly half a point a year. That margin is on an adjusted basis: the forecast puts 2025 gross profit at $571 million, against $492 million under GAAP. Management attributed the margin expansion to "scale and AI-enabled efficiencies, including efficiencies expected from customers added from the acquisition of Enfusion."

Capital spending in the same plan is about 1.5% of revenue every year, mostly capitalized software development, except 2026 at 2.2% for office renovations. The assumptions disclosed in the proxy include no line for consolidating Enfusion's infrastructure, refreshing its servers or running both environments while one moves.

The forecasts also assume Clearwater would continue as a standalone company. That is standard, but it means the buyer's own integration and infrastructure plan is not in the numbers the board relied on.

Two things are being counted on here. The first is AI. The plan credits AI-enabled efficiencies for part of a margin gain worth about $45 million a year of gross profit by 2030. That is a large number to rest on a technology whose own cost, in inference, grows with usage. The proxy says the sponsors' diligence meetings covered generative AI specifically, so the buyers were looking at the same question.

The second is Enfusion. Efficiencies from Enfusion's customers can be real, but they arrive after the integration, and the integration is the expensive part. A plan that books the benefit of the unified platform and none of the cost of building it has its gross margin running ahead of its infrastructure.

The diligence point: where a plan credits AI or an integration with margin expansion, ask for the bridge. Get the cost of each step, including inference cost at plan volume, migration cost and the period of running both environments, and check that it appears somewhere in the plan. If it does not, the margin line is a target, not a forecast.

What a point of gross margin was worth here

PJT Partners, one of the special committee's advisors, selected a range of 18 to 25 times 2026 adjusted EBITDA for Clearwater on a standalone basis. On 2026 plan revenue of $940 million, one point of gross margin is about $9.4 million a year, or about $170 million to $235 million of value at that range. By 2030 a point is worth about $18 million a year.

A migration that holds gross margin even a point or two below plan for two or three years, a mild case next to New Relic's, would have been worth hundreds of millions of dollars in this deal. That is the scale of the question the Enfusion data centers raised.

The questions I would have taken into the data room

1. What does the full infrastructure bill look like across AWS, the Enfusion data centers, dark fiber and every other provider, by month, and which line of the income statement does each sit in?
2. What is the plan for Enfusion's infrastructure: migrate, stay or rebuild? What does each option cost, and which does the forecast assume?
3. What constraints does the Enfusion workload carry, including latency, market data connectivity and client contractual requirements on hosting or data location?
4. When do Enfusion's data center and dark fiber contracts end, what does exiting them cost, and when is its server fleet due for refresh?
5. Which of the $96.1 million in purchase obligations is cloud and hosting, with which provider, and was any of it sized for the combined platform?
6. Which specific AI-enabled efficiencies drive the margin plan, what do they cost to run at plan volume, and how is that inference cost reflected in the forecast?

What the outside view cannot tell you

Plenty. Clearwater does not report hosting separately, its technology costs include more than cloud, and its purchase obligations include software. I do not know why Enfusion chose co-location, what its data center contracts cost, or whether a migration was already planned or underway. Some buyers sat on the board and had a far better view than any public filing gives. Clearwater is private now, and I have no view of what has happened since the close.

None of this says the deal was mispriced. It says the cloud question in the price was visible from public documents before close: one platform on AWS, one in two data centers, a stated goal of unifying them, and a forecast that books the result without the work. That is a short list of questions, and every one of them is answerable in a data room.

Teardowns like this one use only public information. For a target you are actually evaluating, the same analysis is done from the data room, with every infrastructure contract in view. Related reading: When AI-Powered Quietly Breaks Gross Margin on the cost side of AI efficiencies, and the Integration Complexity Scorecard for scoring a platform combination. Book a free discovery call.