Software valuations just hit their lowest level since 2014. Dynamic Business unpacks the Goldman Sachs analysis that shows which companies survive and which ones don’t.
What’s happening: Goldman Sachs has launched a “pairs trade” framework splitting software stocks into two camps: those that will thrive in an AI-driven world, and those it believes face serious disruption.
Why this matters: For SME owners who rely on software tools or hold technology investments, the framework offers a practical lens for assessing which platforms are genuinely built for the AI era and which may be quietly losing ground, regardless of how confidently they market themselves.
Something significant shifted in global technology markets at the start of 2026. Software stocks, which had spent years trading at premium valuations on the promise of predictable, recurring revenues, ran headlong into a question investors could no longer ignore: what happens to subscription software when artificial intelligence can do the same job for less?
The answer, so far, has been a sell-off. According to Goldman Sachs research, the forward price-to-earnings multiple for software has dropped from roughly 35 times earnings in late 2025 to 20 times, its lowest level since 2014 and the smallest premium relative to the broader S&P 500 since 2010. Put simply, investors are paying far less for each dollar of software company profit than they were just months ago.
For most business owners, that might sound like a Wall Street problem. It isn’t. Goldman Sachs responded to the upheaval by doing something unusually practical for a major investment bank. Rather than waiting to see how the dust settled, its US custom baskets team built a structured way to separate the winners from the losers. It is a “pairs trade” that goes long on software companies the bank believes AI cannot easily replace, while shorting those whose core business models are most exposed to automation.
“We expect the long side of the basket to recover from the recent software sell-off, while the short part will lag behind,” wrote Faris Mourad, Vice President of the US Custom Baskets Team at Goldman Sachs, in a note to clients.
The framework sorts companies into two baskets. The buy basket favours software businesses that require physical infrastructure, carry deep regulatory entrenchment, or demand human accountability in ways AI cannot replicate. The sell basket targets companies whose products primarily automate workflows, a function AI is now performing at lower cost.
The logic is more useful than any individual stock recommendation. It gives anyone, from an investor to a business owner evaluating their tech stack, a way to assess software resilience.
Winners and losers
On the buy side, Goldman favours companies serving as the underlying infrastructure of the digital economy. As Dynamic Business has reported, 40 per cent of Australian businesses had adopted AI technologies by the end of 2024, and more than two thirds of Australia’s top technology leaders identified AI as the defining trend of that year. The companies positioned to benefit from that adoption curve are not those selling productivity software. They are the ones providing the pipes, the security, and the data systems that every AI deployment depends on.
Cloudflare, CrowdStrike, Palo Alto Networks, Oracle, and Microsoft are among the names Goldman places on its buy list. What they share is not a marketing message about AI, but a genuine structural position: you cannot run an AI system without compute, cloud infrastructure, cybersecurity, or data storage. Demand for those services rises as AI usage grows, not falls.
On the sell side, the list is revealing. Salesforce, Workday, DocuSign, Atlassian, UiPath, and Accenture all appear, alongside payroll providers, marketing automation platforms, and IT outsourcing firms. What they share is a business model built on automating processes that AI can now describe and perform in plain language.
Goldman is careful not to declare these companies finished. The sell list reflects elevated risk, not certain failure. Some will adapt. But the framework suggests that companies relying on seat-based pricing or workflow automation as their core value proposition face a structural headwind that no product update can resolve.
The picks and shovels lesson
There is a useful analogy from the California Gold Rush. The people who made the most consistent money were not the miners. They were the merchants selling picks, shovels, and supplies to every miner, regardless of whether they struck gold. The same principle applies here. Rather than placing a single bet on which AI company wins the next decade, the more resilient position may be in the infrastructure that every AI system requires to function, such as compute, connectivity, cybersecurity, and data management.
For SME owners, the Goldman framework translates into a practical checklist. The software your business uses or holds as an investment is worth looking at through this lens. Does it require physical infrastructure to operate? Is it embedded in a regulatory process you cannot easily exit? Does it sit in a category, like cybersecurity or data management, where AI adoption increases demand rather than reducing it? If yes, the exposure is likely more resilient than it appears.
If the software primarily automates a workflow, charges per user, and has limited switching costs, those are the characteristics Goldman’s analysis flags as most vulnerable. That does not mean abandoning the product immediately. But it is worth asking whether a more AI-native alternative is already being built to replace it.
The sell-off has pushed software valuations to their most accessible levels in over a decade. Goldman notes that the current multiple is broadly in line with companies growing at 5 to 10% annually, while consensus estimates still project closer to 16 % earnings growth for the sector across 2026 and 2027. Whether that gap represents opportunity or over-optimism is the question the market is now working through.
What Goldman’s framework makes clear is that not all software is facing the same question. Some of it is being disrupted. Some of it is doing the disrupting. Knowing which is which is no longer just useful information for fund managers. It is useful information for anyone running a business in 2026.
This article is for informational purposes only and does not constitute financial, investment, legal, or tax advice. All investing involves risk. Always consult a qualified financial professional before making investment decisions.
Keep up to date with our stories on LinkedIn, Twitter, Facebook and Instagram.
