There’s a special kind of embarrassment in being wrong about something in public, and earlier this year the financial world was wrong about software in public, loudly, all at once. The consensus went something like this: artificial intelligence was going to “obliterate” the software industry. Why would anyone pay for a subscription to a database or an application suite when an AI could just… do it? Software stocks wobbled, analysts wrote elegies, and somewhere a lot of sensible investors quietly sold.
I think about that moment the way I think about all the times I’ve been sure the world had changed forever, only to watch it change back. There’s a humility in markets that keeps humbling the confident.
Because this week, software had its second act. In Axios Markets’ recent editions, reporters Emily Peck and Matt Phillips flagged a striking data point: the State Street Software & Services ETF — a basket of roughly 130 software and services stocks — rose 5.2% on Thursday to a new all-time high. An all-time high. For the industry that was supposed to be obliterated. I don’t know about you, but I find something quietly hopeful in that — not just for investors, but for everyone whose livelihood touches this industry, which is to say, nearly everyone.
And then there was Oracle. Seeking Alpha’s Wall Street Breakfast, in its September 11 edition, led with the kind of earnings report that makes you rethink assumptions. Oracle’s fiscal first quarter: adjusted earnings per share of $1.92 against expectations of $1.75. Revenue of $19.45 billion, up 30% year over year, against an expected $19.13 billion. Cloud revenue of $11.6 billion, up 62%. And then the guidance — the part where a company tells you what it sees coming — pointed to at least $90 billion in revenue for fiscal 2027, above the $89.66 billion consensus, with adjusted EPS of $8.10 against $8.07 expected. Shares jumped 6.3% in premarket trading, after falling 5.3% the day before. A whipsaw, yes, but the direction of the surprise was unmistakable: up.
Let me translate that into something a small business owner would recognise. Imagine you run a mid-sized company — a logistics firm, say, with forty employees. Your software subscriptions have quietly become one of your biggest line items: the database, the planning tools, the customer system. Earlier this year, your nephew told you AI would make all of that free. You waited. And then your Oracle bill went up anyway, and your team kept using the stuff, because it turns out that running a real business on “an AI will handle it” is like running a restaurant on “the customers will cook.” The tools didn’t disappear. They got more embedded.
That’s the story the numbers are telling. Cloud revenue up 62% doesn’t happen because companies are dabbling. It happens because the database is where the company’s memory lives, and nobody — not even in the age of AI — rips out their memory lightly. If anything, the AI boom is making Oracle’s core product more valuable, not less: every AI application needs somewhere to keep its data, somewhere reliable, somewhere governed. The obituary writers confused “AI changes software” with “AI replaces software.” Those are very different sentences.
Now, for the saver watching their 401(k) — and I mean the ordinary saver, the teacher or the nurse who checks their retirement account twice a year and feels a small lurch each time — this matters in a way that goes beyond one company’s earnings. Software stocks are a huge weight in the indexes that carry most retirement savings. When the narrative said software was dying, that lurch was real: the fear that the engine of your retirement was becoming obsolete. A new all-time high for the software ETF doesn’t just repair a portfolio; it repairs a story. The story now is that the digital infrastructure of the economy is not being swept away by AI but being rebuilt around it — and the companies that own the foundations are charging rent.
I want to be honest about the other side, because reflective thinking demands it. Valuations are demanding. A 30% revenue growth rate at Oracle’s scale is extraordinary, and extraordinary has a way of becoming expected, and expected has a way of becoming disappointing. The same AI enthusiasm lifting software stocks today is the same enthusiasm that, elsewhere in the market, is starting to look bubbly — credit markets are beginning to pull back from AI bets even as equity markets double down, a tension I’ll be writing about elsewhere in this edition. Software’s second act is real, but second acts don’t run forever without a third act to follow.
There’s also a quieter human story here about the people inside these companies. When the “AI will obliterate software” narrative peaked earlier this year, I thought about the engineers and support staff and salespeople whose jobs were being narrated as obsolete by people who had never met them. Narratives have casualties before markets correct them — hiring freezes, cancelled projects, anxious teams. A comeback like this one doesn’t just reward shareholders; it tells a few hundred thousand workers that their craft still matters. That counts for something.
Here’s my take, and I’ll label it clearly as opinion: the lesson of software’s 2026 is not really about software. It’s about the rhythm of technological panic. Every platform shift arrives with a prophecy that it will destroy the layer beneath it. The internet was supposed to kill retail; it remade it. Mobile was supposed to kill the PC; it shrank it and the PC survived. AI was supposed to obliterate software; instead, software is becoming the ground AI stands on. The pattern isn’t destruction — it’s absorption. The old thing doesn’t die. It gets a new job.
So where does that leave us? If you’re a business owner staring at a rising software bill, the honest question isn’t “will AI make this free?” but “is this tool earning its keep in an AI world?” — and for a lot of Oracle’s customers, the answer, at 62% cloud growth, is evidently yes. If you’re a saver, the lesson is the oldest one in investing: don’t sell the obituary. Narratives are loud; cash flows are quieter, and they get the last word.
It’s not enough to just chase the newest disruption — we must listen to what customers are actually buying, learn from the forecasts that failed, and contribute our patience to the long, unglamorous work of building things that last. Software was never dead. It was just waiting for its second act. And judging by Thursday’s tape, the audience is back on its feet.
