From AI doom to opportunity: Why Dan Niles has a change of mind on software stocks

This became one of the defining market stories of the year. The Philadelphia Semiconductor Index, or SOX, surged 62%, while the software-focused IGV ETF gained just 4%, trailing the S&P 500’s 13% rise. However, Niles now believes that story may be changing.

In a post on X on August 30, 2026, he said recent market moves and corporate earnings have made him question whether investors became too negative about software. For an investor who has spent decades studying technology cycles, including the dot-com boom and bust, the change in his thinking is notable.

Last wk, SPX/Nas/R2K +0.5%/+0.8%/-1.5% w/ oil -4%. But a hawkish Warsh on Friday led to a bear flattening of the yield curve. Despite $NVDA guide of 70% CY27 rev growth vs consensus of 47%, SOX Index -2.3% while software $IGV +5.9% on solid earnings.

In general, many AI…— Dan Niles (@DanielTNiles) August 30, 2026

The trade has flipped since late June

Niles traces the shift back to June 22, when a major momentum trade began to unravel. He says he had warned about the risks just two days earlier, on June 20. Since then, the market has moved sharply in the opposite direction.

Through August 28, the IGV software ETF had risen 25%, while the SOX index had fallen 22%. Morgan Stanley’s momentum index was down 36% over the same period, while a more concentrated technology, media and telecom trade had dropped 54%.

These numbers suggest that one of the market’s biggest AI bets is beginning to unwind. The trade had been based on a powerful fear that AI would make many traditional software products obsolete. However, the recent reversal has forced investors to reconsider whether they went too far in writing off the sector. Niles is now asking an important question, what if AI does not reduce the need for software at all?

AI agents could actually increase software demand

One of the biggest changes in Niles’ thinking is the possibility that AI agents could become major users of software. Instead of replacing software tools, AI agents could end up using them much more frequently than humans do. He cited estimates suggesting that AI agents could use software 10 to 100 times more often than people.

If that happens, AI could create more demand for software rather than destroy it. This would turn one of Wall Street’s biggest assumptions. The fear earlier this year was that AI would eliminate the need for many software products. However, if AI agents need to access, operate and interact with those tools constantly, software companies could find themselves serving a much larger number of users.

This possibility is making Niles reconsider whether the market was too quick to assume that AI would destroy the value of traditional software companies. Three recent corporate developments appear to have played an important role in that shift.

Three companies that changed his thinking

Atlassian was seen as one of the software companies most vulnerable to disruption from AI. However, after reporting strong earnings and guidance on August 6, its stock jumped 35%. For Niles, the reaction suggested that investors may have underestimated the company’s ability to remain relevant in an AI-driven world.

Then came Workday. The company’s shares rose 18% on August 13 following reports that private equity firm Silver Lake could pursue an acquisition. Niles believes the development “probably put a floor underneath software.” The interest was important because private equity firms usually have to meet high financial hurdles and often rely heavily on leverage when making acquisitions. Their willingness to look at software companies could suggest that valuations have become attractive.

Workday’s earnings on August 27 were also “good enough,” according to Niles. Still, he believes the acquisition speculation probably played a bigger role in the stock’s 6% gain the following day. Salesforce added another important piece to the story. The company gained 23% after reporting results and guidance on August 26. It also announced a new deal with Anthropic, a company Salesforce first invested in back in 2023.

Niles said the partnership would allow users to execute actions directly inside Claude without opening traditional software screens. Salesforce also appears to be changing the way it charges customers, moving towards fees linked more closely to usage and business benefits. These developments made Niles question whether the recent positive reaction to software stocks was simply a market bounce or whether improving fundamentals were beginning to support the sector.

A major change in his long-held view

For years, Niles says he trusted only three areas of software: systems of record, security and gaming. Now, he is wondering whether the market is beginning to recognise a larger opportunity across the software industry. This would represent a major shift from the dominant view earlier in 2026, when investors bet that AI would make many software products obsolete. However, Niles has not suddenly become blindly bullish on everything connected to AI. In fact, his views on artificial intelligence have often been marked by a mix of excitement and caution.

‘100% a bubble’ but not done yet

Niles has previously described AI as “100%” a bubble, but he does not necessarily believe that means the boom is about to collapse. In a CNBC discussion on August 27, 2026, he said he saw little risk of the bubble bursting for at least another year. He remained constructive on Nvidia’s risk-reward but also warning that chip shortages could lead customers to over-order, making demand appear stronger than it really is.

Niles believes the current AI buildout has major similarities with the late-1990s internet boom. Token usage has jumped “10 to 100 times”, while the semiconductor index recorded an exceptionally strong start to 2026. His concern is not that AI is useless but the opposite. The problem, in his view, is that transformative technologies can attract too much money.

“You have 100 companies that each think they’re going to get 50% market share. That’s not going to work,” he told CNBC. That is the core of his bubble argument. AI may transform the economy, but not every company spending heavily on AI infrastructure will become a winner. He previously warned that sometime next year, sectors such as semiconductors could drop 30% to 50%. His biggest worry is the source of the AI money. Even as he rethinks software, Niles remains concerned about the enormous amount of money flowing into AI.

Niles on revenue run-rate of AI companies

Niles in the post also pointed out that the combined annualised revenue run-rate of Anthropic and OpenAI rose from $29 billion at the beginning of the year to $105 billion just seven months later. This kind of growth means companies and customers are spending heavily on AI. One possible source is the roughly $1 trillion global software market outside AI. Another is the IT services industry, which represents about $1.7 trillion in spending and which Niles still believes is vulnerable to disruption.

The biggest pool, however, is knowledge worker pay. Niles estimates that companies spend around $35 trillion to $50 trillion on knowledge workers, representing roughly 30% of the global workforce. If AI disrupts this area, he believes the impact could be “even less noticeable” because the spending pool is so much larger. This is where his thinking becomes more complicated. AI may not destroy software in the way investors once feared, but it could still fundamentally change where companies spend their money.

‘Some real companies are going to go to zero’

Niles has not abandoned his warnings about the software sector entirely. Responding previously to Nvidia CEO Jensen Huang, he told CNBC, “There’s some real companies that are going to go to zero in the software space.” He has said database and cybersecurity companies could be among the most resilient parts of the sector, while more commoditised workflow tools could face a genuine existential threat.

Niles is not saying every software company will survive the AI era. Instead, his latest thinking suggests that investors may have been too broad in assuming that AI would hurt the entire industry. Some companies could disappear. Others could become more valuable because AI agents use their products more frequently. The challenge for investors is figuring out which companies belong in which group.

How Dan Niles became one of tech’s most watched investors

Niles’ influence comes partly from the fact that he has seen this story before. He holds a BS in Systems Engineering from Boston University and an MS in Electrical Engineering from Stanford University. He began his career as an engineer at Digital Equipment Corporation before moving to Wall Street in 1990.

He joined Robertson Stephens’ mergers and acquisitions group and later moved into equity research. Between 1994 and 2003, he became one of Wall Street’s best-known analysts covering semiconductors and computer hardware. He was selected multiple times for Institutional Investor magazine’s All-America Equity Research Team.

In 2000, during the height of the dot-com boom, he was one of only three analysts across all industry sectors to receive three awards covering stock picking, useful and timely calls, and earnings estimates. He also received five Wall Street Journal “Best on the Street” awards. This period, which included both the spectacular rise and collapse of internet stocks, shaped his scepticism about technology buildout cycles.

He later worked at Lehman Brothers and Neuberger Berman before moving into hedge fund management. Today, he runs Niles Investment Management, where he is founder and portfolio manager. His long career has made him a closely watched voice during the AI boom. His latest post suggests that even after decades of studying technology, he is still willing to change his mind when the evidence changes.

Four things he is watching next

Niles also outlined several factors he believes could shape markets in the coming weeks. He said a deal involving Venezuelan oil fields, which he described as holding the largest crude reserves in the world, could help markets start the week positively if oil prices decline. He also believes a Federal Reserve rate hike is likely on September 16, partly because the next meeting on October 28 falls just before the midterm elections.

September is historically the weakest month for markets, according to seasonal trends, and Niles noted that midterm election years can bring even worse seasonality. He is also watching growing bipartisan opposition to data center expansion, which could create pressure for AI infrastructure stocks. Niles said opposition to data centre expansion is one of the few areas where both political parties appear to agree. However, he believes that opposition is wrong and hopes greater public education could change attitudes.

The investor who long believed only a few parts of software were safe is beginning to wonder whether AI could create opportunities across a much wider section of the industry. This does not mean he has stopped worrying about the AI bubble. Far from it. However as the AI trade begins to change, Niles appears to be following his own reminder from Warren Buffett, “the market has to keep pitching but you do not need to swing.”

Disclaimer: This article provides factual analysis only and is not, and should not be construed as, an offer, solicitation, or recommendation to buy or sell securities. Investors must conduct their own independent due diligence and seek advice from a registered financial advisor in the respective jurisdiction.