Key Takeaways

  • MiddleGround Capital mandated Claude across its industrial portfolio, prompting one company to build an internal CRM rather than buying third-party software.
  • Sean Mooney reported that one in three project requests his firm receives from private equity sponsors now focuses on deploying AI into industrial businesses.
  • Middle-market manufacturing acquisitions often run on QuickBooks or fifteen-year-old databases, allowing them to bypass modern enterprise software contracts entirely.
  • John Stewart projects that seat-based software licensing will disappear within five years as language models let non-technical teams generate custom software around proprietary operational data.

Skipping the Traditional Enterprise Stack

Private equity firms buying lower-middle-market industrial companies used to follow an expensive playbook: rip out legacy software, hire systems integrators, and sign multi-year contracts for modern cloud platforms. Stewart says that playbook is dead.

“The average company we buy, they have no like ERP system,” (play from 51:13) Stewart said. “They're using QuickBooks or they're using a system and it's 15 years old and you know they don't have a CRM. You don't need a CRM anymore. I'm sorry Salesforce and you know these other companies you don't need it anymore. If you have your data then you can create all the front end with AI.” (play from 51:18)

Instead of paying millions to deploy enterprise platforms across legacy plants, operators can leave flat relational databases intact. An artificial intelligence layer can query raw tables, draft customer proposals, and track orders without forcing workers into standardized user interfaces. The absence of modern software, once considered a major operational deficit in diligence, now lets industrial targets leapfrog two decades of enterprise software overhead.

Mandating Portfolio Adoption

MiddleGround did not wait for management teams to experiment on their own. Stewart instituted a firmwide mandate requiring portfolio companies to use Anthropic's Claude. The operational result was immediate: one portfolio business wrote its own operational CRM to manage customer workflows rather than contracting a commercial vendor.

“Well, one of our companies, they built their own like CRM system with Claude and don't have to have, you know, a third party CRM system,” (play from 51:47) Stewart noted.

This trend is spreading across the middle market. Mooney noted that sponsor appetite for operations-focused AI tools is accelerating faster than any other operational category. “The fastest growing area that we're getting used by other PE firms is AI. It's growing exponentially,” (play from 50:38) Mooney said. “One in three of the projects that we get demand for are for bringing AI to industrial companies.” (play from 50:43)

Sponsors are using these tools for plant-floor scheduling, machine diagnostics, and pricing decisions rather than hiring software engineering consultants to build custom enterprise applications.

The Collapse of Seat-Based Software

Stewart expects seat licensing to unwind rapidly across enterprise IT budgets. When workers can interact directly with raw operational data through conversational prompts, paying recurring monthly fees for productivity suites and database front ends becomes irrational.

“I mean 5 years from now people don't realize that you're not going to be subscribing to Microsoft Windows,” (play from 41:26) Stewart said. “You're not going to be using Excel and Word and you're not going to be paying that model where they stop selling you software and they start charging you a license fee. That whole thing is going away.” (play from 41:31)

This shift challenges the multiple expansion historically granted to recurring software revenues. “People are looking at industrial companies as a haven now because the software as a service business model is like over,” (play from 41:19) Stewart said. Assets that make physical products have defensible unit economics and physical barriers to entry, while software vendors charging for commoditized front ends face severe pricing pressure.

Why It Matters

This shift signals a major repricing of enterprise value between software and manufacturing assets. Software multiples relied on the assumption that seat-based contracts were permanent enterprise taxes, but generative tools allow businesses to build bespoke software around proprietary data at near-zero cost. Capital allocators are recognizing that physical production holds defensible moats against technical commoditization, while enterprise software vendors face contracting margins and customer defections.