The future of software
Software stocks had a brutal week, aside from a Friday rally, extending what has been a rough year for the industry. I touched base with our tech columnist, Ali Barr, the best I know on AI business models. He also wrote this piece in the midst of the selloff.
Ali, you’ve been covering AI adoption and costs by big business since you started our Tech Memo newsletter. What’s your gut on whether the recent selloff was overblown?
Software business models have underpinned the tech industry for decades. Companies invest heavily upfront to build software, but each additional copy costs almost nothing to distribute. So revenue scales faster than costs, driving fat profit margins. That dynamic helps explain why Microsoft, the world’s largest software company, is so valuable.
AI challenges this model. If this new technology makes employees more productive, companies may need fewer software subscriptions. And as AI tools improve, businesses could replace existing software with AI-driven workflows or even build their own software using AI coding tools. Finally, if software companies embrace AI, that could make their services more expensive to run than traditional software. That would mean rising usage doesn’t automatically translate into soaring profitability.
If software in the AI era becomes less profitable and grows more slowly, then it’s logical that the stock prices of software companies might fall. A lot.
Big Tech spending on AI data centers and other infrastructure is set to soar again this year. How will these companies generate a return on these huge investments?
The numbers are breathtaking. Just two companies, Google and Amazon, are planning capex of almost $400 billion in 2026. A couple more years of the same, and that’s more than $1 trillion.
To get a return on this, they will have to come up with new revenue of well over $1 trillion in future years. AI is amazing and really useful, but it’s hard for some investors to see how this happens. Even if new AI products are awesome, do consumers and companies have enough money to buy all this stuff? I don’t know. One outcome could be that Big Tech giants make do with slimmer profit margins in an AI future. That’s similar to the concerns that have hammered software stocks lately.
Who are the most-interesting people to watch in the sector?
I pay attention to Andrej Karpathy. He was director of AI at Tesla and a founding member of OpenAI. He’s pretty independent nowadays, which means what he says about AI can be trusted more. Bonus: He coined the term “vibe coding.”
Aditya Agarwal was Facebook’s first head of product engineering. He was also CTO and VP of engineering at Dropbox. He’s a coding powerhouse. Recently, he used Claude to do some coding and was stunned by the power of this tool. “I am filled with wonder and also a profound sadness,” he wrote on X. “We will never ever write code by hand again. It doesn’t make any sense to do so. Something I was very good at is now free and abundant.”
I’m usually skeptical, but the start of 2026 feels like a moment of highly disruptive — and destructive — change.
