VIP's 66 layoffs in Vermont: when an acquisition and AI overlap, blaming AI alone is a lazy diagnosis

By WCAX · October 22, 2026.

Vermont Information Processing (VIP), a Colchester, Vermont, company that develops software for beverage companies, laid off 66 workers this week, according to local station WCAX.

Context matters. VIP was recently acquired by a private equity fund with new leadership, which brought large payouts to its employee-owners. Last month it also announced the purchase of Encompass Technologies, touting plans to accelerate artificial intelligence in its technology. The source does not indicate that the company attributes the layoffs to AI.

Mike Battig, a computer science professor at Norwich University who has sent many students to VIP, sums it up: "it's hard to know the interplay between the acquisition and AI, because both are disruptive." In his view, AI is causing one of the biggest disruptions to computing jobs in 40 years. Even so, he notes that the industry recovered in the past with new software engineering jobs and argues that demand for software services and products is growing.

The piece offers another local benchmark: IBM employed 9,000 people at its Essex Junction plant 25 years ago; today GlobalFoundries employs about 3,000, but with far greater productivity thanks to robotics and automation. Labor economist Patty Anderson (Dartmouth) says the economy always adjusts to technological change, but that the key policy question is "how the pie is divided." Senator Bernie Sanders, meanwhile, is calling for binding international rules. Vermont has set up an AI task force to coordinate the public and private sectors.

**Our take**

The thesis: the VIP case is a textbook example of muddled attribution, and that confusion is itself a risk. When a fund buys a company, cost-cutting is practically part of the script. When "AI acceleration" is announced on top of that, the technology narrative can serve either to justify the cuts or to conceal a purely financial decision. With the available data we cannot say which weighed more, and we would be wary of anyone who claims certainty in either direction.

What is defensible is the pattern. In general, software and administrative or entry-level work are the most exposed roles: that is where AI already produces code and processes with fewer people. In the short term, this means a hard transition for the 66 affected workers and for graduates seeking their first tech job. Downplaying that would be dishonest.

At the same time, Battig's and Anderson's argument points to something we share: more productivity means more total capacity, and the GlobalFoundries story shows that far more can be produced with fewer people. The problem is never just how much the pie grows, but who gets it. Here the employee-owners cashed in on the sale, but the workers laid off now have no such safety net: distribution is already being decided, case by case.

In the long run, we maintain that AI can lead us to an abundance in which work is a choice rather than a necessity. But that horizon will not arrive on its own: it requires regions like Vermont to turn their task forces into real retraining, transparency about when a restructuring is driven by AI, and mechanisms to share the gains. A task force is a good start; the test will be whether it arrives in time for those packing up their belongings in Colchester today.