AI won't kill software. It will simply give it new life.

Mint

a day ago

The first song ever played on MTV, which made its debut in August of 1981, was a relatively obscure ditty by The Buggles called 'Video Killed the Radio Star."
The selection was both cheeky and prescient—the new technology transformed the music industry. Album sales declined, but video play soared, eventually leading to the birth of new stars, wider distribution and revenue, and the consolidation of record labels.
Advances in streaming, which developed in the early 2000s despite the industry's effort to stifle it, ultimately added billions more to the pockets of artists, producers, and record companies.
OpenAI could be considered the MTV of artificial intelligence, given how its brand recognition, tech-disrupting ethos, and signature ChatGPT chatbot are sending waves of concern across the software investment landscape.
The latest version, which launched late Thursday, pumped up the volume even more.
GPT-5 which OpenAI calls its most-powerful version to date, can perform tasks such as coding, speech reasoning, and writing. CEO Sam Altman called it a significant step along the path to artificial intelligence.
Jordan Klein, a tech, telecoms, and media sector specialist at Mizhuo, was more blunt.
'It highlighted a lot of new great software coding/design features and added gasoline to the dumpster fire of new AI engines, agents, and co-pilots that are just eating the established front office and [development/IT operations] world," he said.
A lot of that concern is being reflected in some of the biggest enterprise software companies, such as Salesforce, ServiceNow, and Adobe, which rely on sales to larger institutions to power their top and bottom lines.
Salesforce shares are down more than 26% over the past six months, compared with an 8.5% gain for the Nasdaq. ServiceNow and Adobe have slumped 15% and 25%, respectively.
The iShares Expanded Tech-Software Sector ETF, the market's largest, is up around 3.4% over the past six months, but most of the gain is tied to its biggest stocks: Microsoft, Oracle, and Palantir.
A wider view shows that more than three quarters of the ETF's 113 stocks are trailing the S&P 500 so far this year.
Klein at Mizuho worries the bear thesis for software stocks 'will gain more momentum with myriad new AI coding and development tools to come."
Curiously, OpenAI's Altman doesn't see things developing that way, and insisted in an interview with the Economic Times of India that there is no evidence that AI coding tools are replacing software engineers.
'We have badly underestimated how much additional software the world wants," he said. 'AI will make engineers more productive and lower the cost of creating software unlocking huge opportunities for economic growth."
There is some truth to that.
Paul Ashworth, chief North America economist at Capital Economics, points to recent U.S. GDP data which showed a 'massive 86.4% annualized gain in investment in computer equipment and an 18.0% gain in software investment."
But he notes software investment growth is still below average. And he also argues a much bigger surge in software and R&D investment will be needed to translate AI advances into worker productivity.
Bank of America analyst Brad Sills thinks that investment might not be far off, and estimates spending on agentic AI, which performs tasks independent of human supervision, could rise to $155 billion over the next five years.
Interestingly, he likes some of the stocks that have performed so poorly over the past six months, including Salesforce, which markets the Agentforce tool, and ServiceNow, which offers the Now Assist tool.
'We believe industry analysts are underestimating the medium-term uplift to software spending from AI agents," Sills and his team wrote in a recent report.
He also noted it could prove to be a defining catalyst for the monetization of the hundreds of billions hyperscalers are spending on AI-powered data centers.
D.A. Davidson analyst Gil Luria says enterprise software sales are largely based on the number of people a company will need to use it, and the costs tied to either dump it or replace it with a rival.
'Longer term, the answer to the first question could change if AI agents start carving off tasks from employees," he said in a note published Friday.
But that doesn't mean he thinks the software industry is in crisis. Much like the music industry's adaptation in the 80s and the 2000s, he said many software companies are well positioned for the AI revolution.
'While the debate around LLMs moving into the application layer will continue, it is easier to believe that the software that organizes code and data, and then observes and secures it, will be needed even more as AI ramps up," he said.
Beyond companies like Microsoft, which are infusing AI-powered technologies into their suite of software products, Luria likes Snowflake, Datadog, and JFrog.
Video didn't really kill the radio stars. It just created a different platform for them. AI is just as likely to do the same for software.