The engineering case for private equity

Long Lake completed its $6.3 billion acquisition of American Express Global Business Travel on September 29. Co-founder Alex Taubman says it is the firm's 40th acquisition, bringing employment across its companies to nearly 30,000 people. Its plans for Amex GBT include using AI to speed up bookings and help resolve travel disruptions.

Taubman says its first group of acquired companies doubled EBITDA, or earnings before interest, taxes, depreciation and amortization, in less than two years through revenue and productivity growth. Those companies also increased headcount. He says Long Lake has never conducted a layoff. These are reported results from earlier acquisitions; the work at Amex GBT is still ahead.

I think that makes existing service businesses interesting places to invest. A travel-management company, legal-services provider or property-management company already has clients, experienced people, and work arriving every day. An owner can buy into that business and give it the engineering team to rebuild how it operates. The owner can earn a return as the firm serves more customers and offers more to each of them.

Thrive Holdings is pursuing this in accounting and IT services. It keeps practitioners as owners and puts engineers alongside them. Its permanent ownership structure gives those teams time to change processes, train people, and keep improving the business. Thrive reports that one accountant reduced tax-preparation time from 180 hours to 15 using its tools. She used the time to explain returns to clients, add services, and bring in new business.

PE owners already improve operations and share costs across acquisitions. What interests me here is building technology that performs part of the service itself, so a firm can grow without adding delivery staff at the same rate. The buyer funds that engineering; practitioners shape how it works and, where they retain ownership, share in the value it creates.

Ownership gives the team authority to change handoffs and responsibilities around the software. Sequence calls this refounding. Its work with BankSouth included redesigning processes across reporting lines. The company reports reducing loan origination from nine stages to three and application-to-closing time by roughly 52%. Sequence attributes those results to the broader operational redesign, with AI part of the work.

After NetAscendant joined Shield, Thrive's IT group, engineers from Shield and OpenAI worked alongside its technicians in Midland, Texas. Forge, Shield's software, first gathered context before a technician started work on a ticket, then began resolving certain tickets itself. Shield reports that it now handles patch failures that previously competed with customer requests for technicians' time. NetAscendant's founders estimate that the same team can support 25% more users.

OpenAI's Deployment Company puts engineers inside businesses to build production systems. I don't picture a small service firm in the Brazilian countryside hiring that team. NetAscendant's founders say they lacked the staff to build the tools themselves. Joining Shield brought the engineering team to their office.

Consider a hypothetical service firm with $10 million in annual revenue and $2 million in EBITDA. At eight times those earnings, the whole business is valued at $16 million. These are illustrative assumptions, not NetAscendant's financials or market benchmarks.

Suppose it sells enough additional work to increase revenue by 20%, at unchanged prices, while keeping its delivery team and holding the original business's revenue and costs steady. Assume 60% of the extra revenue remains after incremental delivery, sales, and support costs, before the new technology budget. Allocate $500,000 a year for this firm's engineering, software, model usage, and maintenance, including its share of any group costs.

Annual operating economics	Before	After expansion
Revenue	$10m	$12m
Contribution from additional work	$0	+$1.2m
Additional recurring technology cost	$0	−$0.5m
EBITDA	$2m	$2.7m
EBITDA margin	20%	22.5%
Enterprise value at the same 8× EBITDA	$16m	$21.6m

Revenue grows 20%; EBITDA grows 35% after the recurring technology cost. The valuation multiple stays unchanged. If implementation also costs $800,000 upfront, the $700,000 annual earnings improvement covers that amount in about 14 months once fully running, before taxes, capital expenditure, and working-capital needs. The ramp adds time. This is an operating illustration; a shareholder return also depends on financing, cash flows, and the holding period.

At only 10% revenue growth, EBITDA reaches $2.1 million. With no additional sales or cost savings, it falls to $1.5 million. Under these assumptions, the firm needs about $833,000 of additional annual revenue to cover the recurring technology bill.

Bending Spoons shows how this engineering investment can extend across acquisitions. Luca Ferrari describes buying products with established demand and applying the company's strengths in engineering, design, and monetization. Shared tools handle payments, experimentation, and selection among AI models. Other businesses in the group can use those tools, sharing the engineering bill, although each rollout still takes work. The approach predates the current AI wave.

Odevo funds shared technology for its property-management companies, with engineers working alongside property managers. Its German business describes using AI-supported processes for customized reporting. Local teams still need to visit buildings and know their clients; the group finances development that one company would struggle to justify.

Shield reports that Forge reduced the ticket backlog at Exigent, another company in its group. Its engineers now have more time to investigate underlying problems and make proactive updates for customers.