Investing.com -- Industrial software stocks have recently been caught in the broader market sell-off driven by fears over AI disruption, but analysts argue the reaction has been excessive and misunderstands where value is created in enterprise software.

Barclays analyst Guy Hardwick argues the market’s narrative that AI will destroy the economics of software-as-a-service (SaaS) businesses overlooks the importance of services, domain expertise and operational reliability that customers actually pay for.

He said the “SaaS AI bear case is overdone,” adding that AI should represent “an incremental opportunity for our Industrial Software names rather than a threat.” The AI sell-off “is unfairly penalizing Industrial Software,” he emphasized.

The belief AI will commoditize enterprise software is based on the flawed assumption that customers primarily pay for code.

“While AI commoditizes code generation, it highlights the economic value of the service in SaaS. If code is a commodity, the value is the deep domain expertise, reliable service, and hybrid (deterministic + probabilistic) architectures,” Hardwick explained in a note.

“Code is an input, not the product or output,” he stressed, adding that successful SaaS firms moved beyond selling code years ago and now compete primarily on service capabilities and industry expertise.

The analyst said the market’s reaction implies that a large share of enterprise software value comes from coding, which he views as incorrect. Coding represents only a small portion of spending inside mature SaaS companies.

Based on Barclays’ assumptions, coding costs likely account for roughly 4% to 8% of revenue, meaning the direct financial impact from automated coding is limited.

Still, valuations across industrial software names have dropped sharply. Barclays highlights that multiples for companies under its coverage have fallen about 50% in the past six months, with Manhattan Associates (NASDAQ: MANH) shares alone down more than 40% from their 2025 peak.

Another factor supporting the sector is the economics of running AI systems. While generative AI may reduce development costs, operating AI software can be more expensive because of inference costs tied to GPUs, energy and infrastructure.

“Even if software can be generated cheaply or freely, it cannot be operated cheaply at scale,” Hardwick noted, arguing that this creates “a hard economic price floor” that favors established vendors capable of managing those costs efficiently.

Industrial software valuations have fallen back to COVID-era levels and now sit at historical lows relative to the S&P 500, Barclays noted, even as the sector offers stronger free cash flow yields than many industrial tech hardware peers.

Hardwick sees several factors that could support a near-term rotation into the group, including growing fatigue with AI hardware trades, a potential reversal in semiconductor outperformance versus software, and high short interest across the sector.