Who Wins and Who Loses in the AI Wave
It's been exactly 30 months since GPT launched. Now that Google has fully joined the AI race this year with a genuinely formidable Gemini, we've entered an era where AI is simply the new normal — and every small shift it brings tends to have outsized effects.
One interesting development: Google now shows AI summaries at the top of search results — you'll see this the moment you tap into a search on your phone. I suspect Google was reluctant to undercut its own search business by offering AI summaries. But circumstances outweigh intentions: if Google doesn't do it, someone else will. By doing it themselves, they can at least knock a rival like Perplexity down first. As for the lost revenue — well, they'll just have to grit their teeth for now.
Google Joins the AI War, and the SEO Industry Sounds the Alarm
Once Google was forced into this, one industry felt the hit immediately: search engine optimization. SEO is roughly an $80 billion-a-year global industry, and you can picture how the entire sector is essentially tethered to a single target — Google Search, which commands roughly 90% market share. If you're talking specifically about "commercially valuable search," that share might be even higher, since a portion of Bing's search volume comes bundled into the operating system, and you also have to subtract Russia, China, and South Korea — three countries that don't primarily use Google Search.
How many people see the AI summary and just stop scrolling? This phenomenon — getting your answer directly without needing to search further — has a name: "zero-click search." It's actually not a new problem; people were already discussing it back in 2019. But comparing the summary AI of that era to today's is like comparing a child to Superman. For an industry, there's a world of difference between water up to your ankles and water up to your neck — and AI answers and AI summaries have genuinely pushed SEO into neck-deep water.
Another sector taking a serious hit: news and content sites. These outlets rely heavily on search traffic — especially since the good old days of social-media referral traffic ended around 2016, after which news and content sites worked hard not to over-depend on social platforms for distribution. Some outlets that leaned especially hard on social media, like BuzzFeed, ran into serious trouble, and after that, media companies started treating relatively stable search traffic as their crucial lifeline. Now AI has arrived and is telling readers the conclusion outright — people can't even be bothered to glance at the search results anymore, let alone read the full article. And if nobody's reading the content, who's going to click the ads?
Those are just two industries already grappling with the fallout of AI becoming the norm. But beyond these two confirmed casualties, plenty of sectors that people predicted would be devastated have actually turned out fine. The most interesting case, by far, is SaaS (software-as-a-service).
Over the past year or two, countless wise voices — including Microsoft CEO Satya Nadella, who said SaaS would be replaced by AI agents, and Nvidia CEO Jensen Huang, who's repeatedly said everyone will be able to write code in the future — plus the mass layoffs Bay Area engineers suffered in Q1 due to AI, all seemed to point to the same conclusion: SaaS is finished. But the first half of this year has now wrapped up, and things don't look nearly as dire as people predicted for SaaS. Plenty of small and mid-sized AI companies are still running on a SaaS model, and investors keep piling in — there's no sign of things grinding to a halt.
Notably, the well-known VC firm a16z has coined a new phrase: "10x is the new 3x." Top-performing AI companies now grow revenue far faster than traditional SaaS companies ever did. In the past, a strong SaaS company was expected to triple revenue in a year (the so-called T2D3 framework — triple, triple, then double, double, double). But today's best AI companies can grow revenue 10x in a single year.
Plenty of vertical AI SaaS companies are also doing quite well. Sure, a mountain of startups have gone under, but the good AI companies are performing surprisingly well. Cursor, the AI coding tool, has already surpassed $500 million in ARR (annual recurring revenue) — and this company didn't even exist three years ago!
Another case: Oracle, a legacy SaaS company that's pulled off a strong AI transformation, now trades at four times its stock price from five years ago. Before the GPT-driven AI revolution erupted, it rarely even broke $100 a share; now it's above $200. So it's hard to say AI agents or LLMs have knocked SaaS out with a single shot — at minimum, it's premature to draw that conclusion right now, and I think it's still too early to say for sure. Overall, in the short-to-medium term, this has actually been a significant net positive for the industry — it's arguably one of the sectors benefiting from AI, not one being destroyed by it.
AI SaaS Is Heating Up — Mid-Term Outlook Looks Promising
But under the AI shock, "SaaS and software" now takes a hundred different shapes, and how you feel about it depends entirely on where you're standing. I know a friend who left a Big Tech job to start a company amid the AI wave, raised venture funding, burned through it within a year, and had to shut down and go back to job hunting. Fortunately, demand for AI engineers is strong, so he found a new job painlessly. From his vantage point, AI SaaS is a genuinely hard business — that much seems certain.
On the other hand, I've talked with friends at other Big Tech companies where large-scale engineer layoffs are underway explicitly to fund AI — cutting costs to pour money into AI startups and AI services (I suspect that "10x revenue is the new 3x revenue" line is being funded, in part, by exactly this kind of corporate spending). So whether it's actually a good business to be an AI startup is still an open question. But from a founder's perspective, change beats stagnation any day — and over the medium term, I remain quite bullish on where AI SaaS is headed.