Beacon found 160 formulas hiding inside a million

At PIMCO, a spreadsheet with almost a million formulas contained a surprisingly small secret: only 160 of those formulas were unique. Kirat Singh, Beacon’s co-founder, examined sample spreadsheets from the investment manager and mapped their calculations into Beacon’s dependency graph. The structure and functionality remained. The repeated machinery could be handled differently.

It is a useful place to begin a story about financial software. A spreadsheet can be a remarkably successful prototype. Its inconvenience arrives later, when the same logic must serve more people, larger portfolios and more scenarios. PIMCO wanted a platform better suited to scenario analysis and stress testing. Beacon’s answer became Excel Evaluator, a tool for turning spreadsheet logic into backend models.

The interesting unit here is the unique formula. A million cells looks like a million problems; repetition changes the diagnosis. Beacon’s case study describes a transformation of representation, not a million-to-160 speed benchmark. The lesson is more practical: look inside the thing you intend to replace before discarding what it knows.

Two men who had built this before

Beacon began in 2014 with Singh and Mark Higgins. Their experience ran through the institutional platforms that financial engineers know by short, almost domestic names: Goldman Sachs’ SecDB, JPMorgan’s Athena and Bank of America Merrill Lynch’s Quartz. These systems connected financial models to the machinery needed to use them across a bank.

In an interview, Singh described realizing that he was doing the same kind of work at different institutions because a suitable external platform did not exist. Beacon commercialized that experience. The company licenses developer infrastructure and financial applications, giving customers a foundation they can modify for their own purposes.

This is a particular answer to the perennial buy-or-build argument. Buying saves the work of constructing common infrastructure. Building preserves the institution’s peculiar expertise. Beacon sells the common foundation and invites clients to extend the models. A bank’s advantage need not reside in maintaining another computing framework.

Blackstone’s problem was waiting

Blackstone had built a centralized data platform with Snowflake. Its analytics still lived across spreadsheets and third-party tools. According to its published case study, spreadsheets struggled with the scale of its data and scenarios, while external risk tools lacked the transparency needed for customization. Consistency and reproducibility were difficult to enforce.

Consider its credit analytics. When calculations became too complex for spreadsheets, teams requested purpose-built changes from a vendor, waited for testing and delivery, then waited for the analysis. Each result might suggest another adjustment. Curiosity had acquired a queue.

With Beacon, the case study reports, teams could adjust models internally and run calculations across tens of thousands of positions in minutes. Snowflake supplied the shared data; Beacon supplied a shared environment for models and computation. Other teams could reuse the same models. The improvement was partly computational and partly organizational: fewer handoffs between a question and its next version.

“Data without analytics is noise, and analytics without data is useless.”Adam Lichtenstein / Blackstone

The plumbing has a Python address

The product serves quants, developers, traders and risk managers. Its applications support pre-trade pricing, portfolio views, scenario analysis and risk reporting. Beacon Notebook gives researchers an environment for experiments; development workflows provide review and version control as those experiments move toward production. Existing spreadsheets can remain a familiar interface.

Its transparent source-code license is central to the bargain. Clients can inspect and extend commercial code, add instruments and adapt pricing or risk analytics. That permission belongs to a licensing relationship. Buyers should establish precisely what they can modify, operate and retain under their contract.

Beacon also preserves histories of trades and data, including when an event applied and when it entered the system. A correction can therefore be examined without erasing the earlier picture. For a risk team trying to reproduce yesterday’s result, those two clocks matter. A handsome dashboard is of limited assistance if nobody can explain why its number changed.

The same machinery can price power

The customer list reaches beyond investment desks. CleanChoice Energy needed to manage residential electricity demand, hedge purchased power and analyze long-term power-purchase agreements. Its proprietary models required inputs updated manually several times a day, while different forecasting horizons demanded substantial processing capacity.

Working with Beacon’s commodity experts and quantitative developers, CleanChoice built an energy valuation engine, a sensitivity analysis tool and audience-specific risk dashboards. The models incorporated such inputs as wholesale power costs, customer migration and infrastructure charges. Here, the useful output was a defensible electricity price and a clearer view of exposure.

That example explains Beacon’s position in the market. It is an enterprise platform for institutions with complicated instruments and their own modeling requirements. The closest alternative may be an internal development project, a spreadsheet estate or a specialist trading-and-risk system. A team with simple reporting needs would have less reason to assume the work of customization.

A platform acquires a larger address

Customers also helped finance the business. PIMCO led its 2018 Series A; Centana led a $20 million Series B announced in January 2020. A $56 million Series C followed in October 2021, led by Warburg Pincus and joined by Blackstone and existing investors. The people buying the machinery were sometimes buying shares in its maker.

In March 2025, Clearwater announced a Beacon purchase price of approximately $560 million, with 60% in cash and the balance in shares under the announced terms. It reported roughly $44 million in Beacon annual recurring revenue at the end of 2024. ARR describes recurring contracts; it should not be substituted for annual recognized revenue.

The deal closed on April 30. Clearwater’s subsequent interim accounting recorded $531.973 million in total merger consideration at closing, reflecting the accounting valuation rather than simply repeating the announcement. Clearwater positioned Beacon alongside Enfusion and Blackstone’s Bistro software to connect modeling and trading with accounting and reporting.

For a prospective customer, the lesson is to begin with one stubborn workflow and test it using actual models, data and reporting requirements. Beacon’s examples show the value of keeping useful logic while changing how it travels. That approach needs technical ownership and governance inside the institution. Someone must still decide which assumptions deserve to survive the move.