It has been a difficult start to 2026 for the software industry. New artificial intelligence (AI) product releases from the likes of Anthropic have investors questioning whether AI agents will replace or merely buy fewer software licenses.
What's interesting is that two of the most bullish technology analysts on Wall Street, Tom Lee and Dan Ives, appear to have opposite takes on the sell-off: one thinks the software disruption is real, and the other calls the software meltdown a "golden buying opportunity." So who is right?
Image source: Getty Images.
Disruption, or "DeepSeek" moment?
Some investors may recall that one year ago, the entire AI complex sold off when Chinese AI lab DeepSeek released an extremely efficient large language model (LLM) that cost a fraction of what other models cost to train.
However, that panic-sell turned out to be a golden buying opportunity. Thanks to "Jevon's Paradox," the idea that use of something increases as it gets cheaper, AI investment has only accelerated from that point.
Could the same thing happen here with software? While the S&P 500 is about flat for the year, the largest software exchange-traded fund, the iShares Expanded Tech-Software Sector ETF (IGV 0.88%), is down a nauseating 21.7%.
The sudden and widespread sell-off seems to have come on the back of releases of increasingly proficient artificial intelligence applications, with the worry accelerating into February with the release of Anthropic's Opus 4.6 model and related applications Claude Code and Claude Cowork.
One prominent feature of Opus 4.6 is that it enables agent "swarms" for Cowork. Or, instead of one agent doing many different things one after the other, 4.6 enables agent "teams" that can divide up work to do them simultaneously on a computer.
This is just one of many rapid improvements Anthropic and other models have made over the past few months, and it's causing a lot of consternation in traditional enterprise software stocks. After all, if AI agents are now able to write software and perform all of these functions that humans recently performed with software as their primary tool, does the importance and value of the underlying software go down?
NYSEMKT: IGV
Key Data Points
Tech bulls divided
What's interesting is that this AI wave is so rapid, and the software stocks have declined so much, that even the more bullish tech analysts on Wall Street are divided. Fundstrat's Tom Lee recently said in an interview that "it's not lost on the Fed that AI is wreaking havoc across software now, and that means that job losses are soon to follow." Of note, Lee has been one of the more bullish analysts on Wall Street, especially when it comes to tech.
On the other hand, Wedbush sell-side analyst Dan Ives chimed in differently, saying, "In our opinion, in my career, going back to the late Nineties, it is the most head scratching sell-off relative to what I believe is going to be the opposite as it plays out that I've ever seen." Ives believes that AI use cases will explode so much that overall usage of critical software will increase by leaps and bounds, making up for any pricing pressure.
These two perspectives might not be 180 degrees from each other, and differences may lie in the details.
What executives are saying
For the record, many tech executives say today's fears are overblown; however, many tech CEOs also have an incentive to say that. Nvidia (NVDA 1.53%) CEO Jensen Huang recently said that AI will probably use software tools rather than inventing entirely new ones. And that, of course, is likely to be the case. Why build a wholly new tool when AIs can merely use what already exists?
But mere usage isn't the full story when it comes to these stocks. There is also an argument that since AI agents will be using tools in this scenario, software companies won't be able to sell as many licenses, or "seats," to large corporations. But countering that notion, in a recent CNBC interview, ServiceNow (NOW 2.30%) CEO Bill McDermott said that if that scenario happens, ServiceNow has a hybrid pricing structure that can operate either on a seat-based or usage-based model, and it's up to customers to choose.
Meanwhile, software companies appear to be falling over themselves to partner with and integrate with LLMs to deliver tangible outcomes. ServiceNow and Figma (FIG 5.82%) both recently announced partnerships with Anthropic, while Salesforce (CRM +2.36%), another software giant caught up in the downdraft, inked a deep partnership with OpenAI in October of last year.
But if these software companies integrate LLMs for a large part of their future value, do the LLMs get a bigger piece of the pie? Can software companies charge what they've been charging if a lot of the value comes from a third-party LLM?
How does this shake out?
While some software companies may come out the other side of this OK, investors should prepare for a period of disruption and uncertainty. It also doesn't help that software stocks have generally traded at higher valuations than most other stocks over the past decade or so coming into this moment, given their "asset-light" business models and strong growth prospects.
One possible bulwark is that legacy enterprise software providers such as Salesforce and ServiceNow have a wealth of data from large enterprises in their systems dating back years or even decades. If a newcomer developed its own AI-generated software with similar capabilities, would it be able to persuade a large enterprise to rip out all its data and migrate it to a new, unproven upstart? Or would the incumbency and trust of the older platform win out?
As it stands, I'd bet that the incumbents figure out a way to survive. But the certainty on this assumption is low. If a newcomer comes along with similar capability at one-tenth or even one-twentieth the price or less because building software is so much easier, would that be enough to persuade enterprises to switch? At the very least, could customers demand lower prices? However, would usage increase so much that overall dollars would grow even if the price per token or price per instance came down?
These answers won't be answered for a while. As such, I'd expect the valuations of major software companies that aren't building their own LLMs to remain at lower valuations than their recent history. And while this sell-off could make for a great long-term buying opportunity, I don't think a rerating higher is going to happen anytime soon.
