AI-native. Series B-backed. Gone in three years.
Fashionable and well funded doesn't equal durable.
The .ai land grab
Open your inbox. Scroll through the vendor pitches that have landed in the last six months.
Count how many have a .ai domain. Count how many claim to be “the first AI-native [something] platform” or position themselves as the “AI-first” solution in their niche.
This is the current state of B2B software positioning, and it’s working.
Deals are getting done. They’re genuinely impressive in a demo, they move fast, and they’re being funded aggressively.
Fast growing businesses are adopting these new-breed tools and attempting to run critical operations on them.
But there’s a big, unspoken risk attached to such a move…
There are going to be far more venture backed businesses with .ai domains that fail in the next five years than succeed. That’s just how venture maths works at scale. And right now, fast-growing retail brands are considering building their operational backbone around these tools as if that reality doesn’t apply to them.
But it does.
The numbers behind the narrative
BLS (Bureau of Labor Statistics) data shows that 20% of newly created businesses don’t survive their first year. 50% are gone within five. Only 34.7% of businesses born in 2013 were still operating a decade later.
That’s not software-specific. But before you sign a multi-year contract with a company that was incorporated eighteen months ago and has a founder who’s never shipped an enterprise product before, it’s the right place to start.
The instinct is to dismiss this. “We’re not buying from some seed-stage startup. Our vendors have raised serious money.” That’s where the thinking tends to go wrong.
Venture economics are built on extreme skew. Only 1 in 140 VC-backed companies ever becomes a unicorn. The model assumes a small minority of portfolio companies become massive winners and a large majority don’t. That maths works fine for a fund managing across dozens of bets. It’s a terrible basis for choosing the system that runs your inventory, your finance, or your order orchestration.
Fashionable and well-funded is not the same as durable. The more capital floods into a single narrative, the more companies that narrative attracts - including a lot that won’t survive the consolidation that always follows a hype cycle. Funding is not a durability signal. It’s a signal that someone made a calculated bet on upside. Those are different things, and the retail industry keeps confusing them.
The question missing in the evaluation room
Being first to adopt an emergent tool in your marketing stack is a calculated risk worth taking. Being first to run your inventory, your finance, and your fulfilment on one is a different kind of bet entirely.
If it goes wrong, you’re not dealing with a campaign that underperformed. You’re dealing with an operational crisis, a migration under pressure, and a vendor that may not have the resources to support you through either.
The vendor evaluation process at most brands goes the same way. Features. Pricing. Integration complexity. Maybe a reference call with a brand two sizes bigger than yours. Nobody asks whether the company will still exist in three years. Nobody models what a forced migration would cost. Nobody asks what happens to their data if the Series B doesn’t close.
Those are the questions that should be leading the conversation. Especially now.
Private equity isn’t necessarily the safe alternative you think it is
The counter-argument usually goes: “We don’t buy from scrappy startups. Our vendors are PE-backed.”
Moody’s data shows PE-backed companies defaulted at roughly double the rate of non-PE-backed peers between 2022 and 2024. Even among the largest, most established sponsors, default rates sat around 14%. S&P counted over 80 US PE portfolio company bankruptcies in 2024 alone.
What retail operators should actually be doing
The problem isn’t that brands are buying software from young or PE-backed companies. Some of the best tooling available right now comes from exactly those places. The problem is that the evaluation process doesn’t match the stakes.
If a vendor sits in the control layer of your business - finance, inventory, order management, replenishment - the bar for selection and ongoing management should be different from a tool running your email campaigns or your content.
For the control layer, the questions that actually matter:
On operational continuity. If this vendor ceased trading tomorrow, what’s your data extraction plan? How long would a migration take, and what would it cost? Have you modelled that scenario, or is it just uncomfortable to think about?
On financial durability. When did they last raise, and what does their runway look like? What’s their path to profitability?
Is it really worth the risk. When compared to established vendors in the space, do the feature and speed benefits really justify the inherent risks?
Most brands can’t answer those questions for their current vendors. That’s the real exposure.
The contrarian position
The standard advice is to avoid young vendors and stick to established platforms. That’s not quite right either.
The right frame is: match vendor maturity to workflow criticality. New, AI-native tooling can earn a place in your stack. Just maybe don’t put it at the centre of your entire operation just yet. Run your experimentation, your personalisation, your content tooling on the exciting new thing. Run your finance and inventory on something with a 15-year track record, a deep implementation ecosystem, and a balance sheet that’s survived at least one economic downturn.
The brands getting this wrong aren’t naive or foolish. They’re running an evaluation process that assumes software companies to be more durable than the data suggests they actually are.
Reliability is about to get expensive again. The vendors that survive the next round of consolidation will charge more for it, and buyers will pay, because they’ll have learned the hard way what the alternative looks like.
The uncomfortable truth
Somewhere right now, a founder is in a board meeting celebrating a new ERP go-live. The vendor they chose has eighteen months of runway and a Series B that hasn't closed. Nobody in that room knows. That's the problem.