The problem with “sell before AI wrecks your business”

The latest argument making the rounds in some forums I have seen is: “AI is about to decimate almost every industry, and the smart move is to sell now, before the market catches on and your multiple falls off a cliff”. One author compares it to a tsunami warning, advising founders to “sell before the wave hits. Wait too long and you are left holding the last Blockbuster on the block.”

Like most arguments there is some truth buried in the approach. Some businesses are genuinely exposed. If your entire value proposition is a repetitive task that a language model now does for a tenth the cost, you have a real problem, and you should be thinking hard about your next move. I have sat across the table from owners in exactly that spot, and my advice to them has been blunt.

But “AI is going to decimate most businesses in almost every industry” is a big claim, and it falls apart once you look at what is actually happening in the deal market. From where I sit, running both buy-side and sell-side processes, the story is a lot more interesting. And for a good number of owners, a lot more encouraging.

AI is destroying some businesses. It is making others worth more.

Here is an example straight from my own desk. I am currently running a sale process for a company that provides web data infrastructure — the plumbing that lets other businesses collect and use data at scale. Two years ago, the company was pitched as growing steadily but unglamorous. Today it is drawing serious inbound interest specifically because companies, especially AI builders, need exactly what it sells: clean data, at scale, reliably. AI is not a threat to that business. AI is the reason multiple buyers are competing for it right now.

That is not a one-off. Look at the physical, hands-on side of the economy — the part these arguments assume more protected from AI. The theory goes that, mostly by accident, these companies are harder to disrupt because of the human labor needs.

AI turned what looked like a weakness into a reason to pay more. Multiply that across HVAC, electrical, cooling, and power businesses tied to the data center buildout, and you have a wide slice of ordinary, unsexy companies where AI is pushing multiples up, not down. This represents the new opportunities and does not begin to highlight how these businesses are getting more efficient using technology.

We have run this movie before

If “AI eats everything” feels overwhelming, it is worth remembering how many times smart people made the identical claim about a different technology and got it wrong.

Eric Vishria, a general partner at Benchmark who has backed companies like Confluent and Cerebras, points back to 2014, when the consensus among venture investors was that Amazon Web Services would swallow the entire software industry. Cloud would undercut every SaaS company on price, since AWS could offer infrastructure at a fraction of the margin those companies needed to survive. Jeff Bezos’s own framing became the industry mantra: a fat competitor’s margin was simply his opportunity.

It did not play out that way. Snowflake built a business competing directly with Amazon’s own Redshift, running on Amazon’s own cloud, and it worked. Confluent, Elastic, MongoDB, and Databricks all built durable, high-margin companies in the years everyone assumed AWS would own. Azure and GCP, written off as irrelevant at the time, became two of the most valuable franchises in the world.

“It feels like we’re going to end up with an oligopoly of winners.”

Eric Vishria · Benchmark

That does not mean nobody loses. Companies that did not adapt to the cloud shift got hurt, the same way some companies will get hurt by AI. But “some businesses lose, others adapt and grow, and a few new giants emerge” is a very different prediction than “AI decimates most businesses in almost every industry.” One of those predictions has already played out once, in the exact market everyone keeps comparing AI to.

There is a second example worth knowing. Vishria points to Geoffrey Hinton, who said in 2016 that hospitals should stop training radiologists because AI would soon read scans better than any human. Hinton was not wrong about the underlying capability. He was wrong about how fast that capability would replace anyone in the real world. The training data did not exist in usable form. Reimbursement, liability, and malpractice law were all built around a human signing off. And once AI did start helping, cheaper imaging led to more imaging, and radiologists ended up in higher demand, not lower.

That gap — between what a technology can theoretically do and how fast the real world absorbs it — is exactly what the tsunami argument skips over. It is also exactly where a well-run business, or a well-timed sale, actually happens.

Worth noting too: Benchmark itself just raised a growth fund for the first time in years, specifically to chase a small number of highly differentiated companies at larger check sizes, not to spray capital across anything AI-flavored. That is not the posture of a firm that thinks a wave is about to wipe out the board. It is the posture of a firm that thinks the winners will be fewer, bigger, and worth paying up for — a very different bet than “get out now.”

If AI is really coming for your industry, the buyer already knows it

Here is where the tsunami argument breaks down completely. The whole thesis depends on a gap: sellers who understand the AI risk, and buyers who do not, at least not yet. It assumes only the sellers can see the wave and feel the earth shake. It tells you to sell into that gap while it is open, and collect a price built on a buyer’s blind spot.

That gap does not exist in any deal process I have run or watched in the last two years. Buyers — especially private equity and strategic acquirers — are running AI-specific diligence on every deal that crosses their desk.

Baker McKenzie’s 2026 dealmaking outlook calls today’s buyers the most disciplined and well-informed group the market has seen in years. I agree. PwC’s mid-2026 outlook shows software acquisitions built around AI capability losing steam against last year, with valuations coming down specifically where buyers think AI threatens the underlying revenue model. Buyers are not sleepwalking into disrupted businesses. They are avoiding them, or pricing them for exactly what they are.

So ask the question the original thesis never asks: if AI is genuinely coming for your industry, who is going to pay you top dollar for it? A buyer who has not done their homework? Good luck finding one in this market.

Markets need two sides, and right now sellers are the weaker one

Here is the part that actually concerns me, and it is the opposite problem from “sell now or get nothing.” I am not seeing sellers who understand AI too well and need to rush for the exit. I am seeing sellers who do not understand how AI touches their own business at all, in either direction. They cannot articulate their data moat. They cannot explain why their workflow has real switching costs. They walk into a process assuming AI is either irrelevant to them or a death sentence, and neither framing gets them a good outcome.

Buyers who have done the actual work — who know where AI strengthens a business and where it erodes one — are capturing the value in this market. They know a proprietary dataset is worth more today than it was three years ago. They know a business with entrenched workflows holds up fine against a chatbot. They know which physical, labor-heavy businesses just picked up a tailwind and which ones are exposed. And they price accordingly. Sometimes generously, sometimes not, but always with information most sellers have not bothered to gather about their own company.

That asymmetry — an informed buyer against an uninformed seller — is the real risk in this market. Not a tsunami. A pricing gap that runs the wrong direction for owners who have not done the work to understand where they stand.

So what should you do

Forget the tsunami. Do the diligence on your own business that a serious buyer is already planning to do on it.

Some businesses should sell now, while AI-driven demand for what they do is running at a peak that will not last forever. Others should hold, invest in the moat AI just made more valuable, and sell into an even better market in two or three years.

Either way, the decision should come from understanding your specific position, not from a blanket theory that AI is coming for everyone at once. It is not. It is coming for some businesses and building others up. Your job, and mine, is figuring out which one you are actually running.