On 30 June, Vontier closed the sale of Teletrac Navman to the private equity firm Respida Capital. The headline valuation was 220 million dollars, which in our industry is a real transaction and not a footnote.
The structure is where it gets interesting, and you have to go to the quarterly filing to find it. Of the consideration Vontier actually received, worth roughly 208 million dollars by its own estimate, 85 million came as cash at closing. Most of the remainder took the form of seller’s notes, which is to say that Vontier lent the buyer a large part of the purchase price. A further 23 million was a minority stake Vontier kept in the business it was selling. And Vontier recorded a preliminary loss of about 86 million dollars on the transaction. It is all in Vontier’s quarterly report, under divestitures.
So the seller financed a substantial share of its own exit, held on to equity in the asset it was trying to leave, and wrote off 86 million dollars for the privilege.
Does that sound like a great deal to you? No? It is the benchmark for 2026 all the same, because it is one of the very few transactions in our space that actually landed this year.
Where We Were Twelve Months Ago
For most of the last fifteen years, consolidation was the defining activity in fleet management. Many small players, healthy growth, PE-backed companies pushing toward the next round, and a steady supply of buyers who believed that assembling FMS companies across geographies under one roof would produce a multiple worth more than the sum of its parts. Quite often without much investment in actual synergies afterwards.
Last year still looked like that. ECM sold YellowFox to Peter Möhrle Holding after roughly five years of ownership. Geotab took over Verizon Connect’s commercial operations across nine countries in Europe and Australia, a transaction worth pausing on, because Geotab bought the salesforce and the customer relationships and explicitly left the product and engineering teams behind. Geotab paid for distribution and declined the software.
And this year? One transaction of any real scale, structured as described above. There have been smaller deals. I also know of several quiet attempts to place assets, most of them shown to a deliberately narrow audience so that a failed process does not leave a mark on the asset. Processes have always failed, that is nothing new. But eight months into the year with a single significant closing, and that one part-financed by the seller, tells you something about where buyers and sellers currently disagree.
The reason has a two-letter abbreviation: AI.
SaaSpocalypse: What the Market Actually Repriced
In January, Anthropic launched Claude Cowork. Within weeks, something in the order of a trillion dollars of enterprise software market capitalisation was gone. The financial press called it the SaaSpocalypse, and Oliver Wyman has published a good analysis of what it did to software valuations.
Before arguing about whether the discount is deserved, it helps to be precise about what was repriced, because two separate arguments got compressed into one headline.
The first is seat compression. If AI agents do the work of ten people, nobody pays for ten seats, and every business whose revenue scales with headcount loses part of its base.
The second is substitution, and it did more damage. The clearest illustration came from CNBC, where two reporters with no coding background set out to build a replacement for Monday.com using Anthropic’s tools, and had a working version about an hour later. Monday.com lost roughly 37 percent of its value over the course of February, with the sell-off starting in the first week of the month and continuing through a disappointing earnings report. The argument is that companies will stop paying for a platform that fits eighty percent of their requirements once building the twenty percent they actually need has become cheap.
Here is the part that gets overlooked. This was not a case of the M&A market reacting to a public market panic. Bain’s 2026 M&A Report, a survey of more than 300 M&A executives published in January, found that one in five strategic dealmakers had already walked away from a deal because of AI’s anticipated impact on the target. That was written before the February selloff. Buyers were pricing this in while the public markets were still comfortable.
Which means the discount now being applied to our industry reflects a diligence question that professional acquirers had already started asking, rather than a sentiment problem that will pass when the news cycle moves on. Whether it is the right question for fleet management is a different matter, and that is what I want to work through.
Three Reasons Fleet Management Is Different
Per-Vehicle Revenue Does Not Compress With Headcount
Take the seat compression argument first, because it runs out of road quickly here.
Rough numbers from my own years in the industry: a 40-tonne truck is a €120,000 asset that burns €30,000 to €50,000 of fuel a year and requires a driver costing €50,000 to €70,000. Against those figures, a one or two percent improvement in utilisation or fuel consumption pays for the subscription several times over. That arithmetic sits at the asset level, and fleet management has always been priced accordingly, per vehicle rather than per user.
Agentic AI does not reduce the number of trucks a haulier operates. It may well reduce the number of dispatchers, planners and back-office staff logging into the system, and for a seat-priced product that would be the whole ballgame, although I would argue that even this is slower than the market assumes, for reasons I set out in Why Agentic AI Hasn’t Taken Over the Dispatch Desk (Yet). Here it changes very little about what the customer pays. A fleet that automates half its planning function still runs the same 200 vehicles and still pays for 200 vehicles.
The mechanism that drove the February selloff has no clean transmission path into this revenue model. An acquirer applying a blanket software discount on seat-compression logic is pricing in a risk the asset does not carry.
Churn Is a Workshop Visit
In normal SaaS, losing a customer is a cancelled subscription. In fleet management, losing a customer is a small logistics project.
Every customer relationship comes with installed devices in vehicles, each with a physical lifecycle of five to ten years, each pointing at a specific backend. Replacing a provider means getting several hundred vehicles into workshop bays, retraining drivers, rebuilding API integrations that the customer’s own IT department or a third party controls, and migrating years of historical data that legal retention requirements say must come along.
I have run four platform integrations at Webfleet. Work of this kind takes three years when it goes well, and the timeline is set by the customer rather than by the provider. When a vehicle can be brought in for hardware work, when an API integration can be changed, when drivers can be retrained: none of these are decisions the software vendor gets to make. That deserves an article of its own, and I will write one, because the gap between what a migration costs on paper and what it costs in practice is one of the most underestimated numbers in this industry.
The result is customer relationships that routinely run eight to fifteen years. Switching costs here are physical, and cheaper software does not change the cost of a technician’s time.
The Liability Attaches to a Device Somebody Can Inspect
A significant share of what fleet management systems sell is regulatory. Tachograph handling, driving time compliance, the protection of driver data (which in Germany also means works council agreements), and the growing pile of emissions and sustainability reporting obligations.
Plenty of vertical software carries regulatory weight, so on its own this is not much of an argument. What makes it different here is where the liability attaches. A driving time record came off a specific piece of certified hardware, in a specific vehicle, at a specific time, and an authority can go and inspect that vehicle. Writing the parsing logic is cheap and getting cheaper. Being the party that is certified, insured and still solvent when that record is questioned three years later is a different proposition entirely.
Every new mandate of the past decade has increased demand for telematics rather than threatening it. Regulation is a tailwind in this industry, and it accrues specifically to providers who can carry the consequence.
Three Reasons the Discount Is Deserved Anyway
I have sat on the diligence side of enough transactions to know that the comfortable argument is rarely the complete one. Here is the case against everything above.
These Were Never Software Margins
Fleet management businesses typically run gross margins of 55 to 70 percent against 80 percent and above for pure software, because of hardware COGS, M2M connectivity and field service. Working capital sits in inventory. There are workshops, installer networks, RMA logistics and spare parts behind the subscription line.
And in a good number of these businesses, the hardware is not sold to the customer at all. It is rented. Which means part of what a buyer is being asked to value on a software multiple is, in plain language, a leasing book: devices sitting on a balance sheet, depreciating on a schedule, throwing off a monthly fee. There is nothing wrong with that business and it produces beautifully predictable cash. But nobody has ever paid a software multiple for a leasing book, and describing one in SaaS vocabulary for long enough that everybody stopped noticing does not change what it is.
Some of the current correction was overdue before anyone had heard of an agentic workflow. A seller who attributes the whole gap to AI is telling themselves a story.
The Threat Is Downgrade, Not Departure
At Webfleet we had a customer in public transport who worked with us for years, decided to leave, and built their own fleet management system. They came back when cameras entered the market, I would guess because at that point the cost of duplicating everything became visible. Building around a provider who handles the ugly parts, hardware sourcing, connectivity, CAN data management, certification, remains worth it for most organisations even now.
So the full-scale rebuild is still rare. What has changed is the middle path.
Every provider has APIs. Consuming a cheap, basic tier and building a thin layer of company-specific software on top of it is entirely viable today, and for an organisation with a competent IT function it is often the right answer. They get the parts that are genuinely hard to replicate and they build the parts that are specific to how they work, integrated properly into their own landscape.
For the provider, this is worse than losing the customer, because it does not look like losing the customer. The logo stays. The contract renews. Average revenue per vehicle falls, the subscription mix drifts from high-value tiers to low-value ones, and the growth story degrades while every retention metric on the dashboard looks healthy.
The Buy-and-Build Thesis Is Broken
For fifteen years, the standard play in this industry was to acquire across geographies and promise platform synergies. I have seen cases where more than half of an acquired company’s installed base had churned out before the migration was finished.
Think about what that means for the multiple. The buyer paid for subscribers and inherited a migration that outlived a meaningful share of them. The synergy model assumed a platform consolidation that arrived, when it arrived at all, years late and considerably smaller than acquired.
A sponsor buying into this space today has to underwrite an exit to someone who believes in year ten. That is a harder story to tell than it was in 2021, and it has very little to do with whether an AI agent can write a dispatch report.
What I Would Actually Look At
Uniform discounts applied to non-uniform exposure create both bargains and traps. The filings will not tell you which one you are looking at. I have written before about why technology due diligence so often misses the point, and the current environment makes that failure mode more expensive rather than less. Here is where I would spend diligence time.
Split revenue by what protects it. Physical protection (installed hardware base, workshop-dependent switching). Contractual protection (multi-year terms, hardware financing). Regulatory protection (compliance-driven, liability-bearing). And unprotected revenue: reporting, dashboards, basic workflow. That last bucket is where the software discount genuinely belongs, and its size varies enormously between two companies with identical top lines.
This matters more now that the replication test has become standard practice. Bain has started building AI replicas of target products during diligence, an approach it calls outside-in diligence, and at least one investor has walked away from a bidding war after seeing how quickly a target’s analytics platform could be reproduced. Apply that test to a fleet management business and you get a useful split. The interface, the reports and the dashboards can be replicated in days. The installed base, the certifications and the liability position cannot be replicated at all. Any diligence process that stops at the first half is measuring the wrong thing.
Look at usage depth, not licence count. My estimate from twenty years in this industry is that roughly half of fleet management customers use the map and perhaps three reports. No serious dashboards, no advanced reporting, no sophisticated geofencing or alert configuration. That is a rule of thumb rather than a measured figure, but the shape of it holds, and I have written about the same pattern from the product side in The Agentic AI That Already Works in Your Fleet. Those customers are easy to lose. What makes customers stay is depth: API integrations into their own systems, driver scoring and coaching, tachograph data management, order management with or without a TMS. At Webfleet we built churn predictions on usage patterns and they worked. Any target should be able to produce this analysis. If they cannot, that is itself a finding.
Track ARPU and the subscription mix over time, not just in aggregate. The downgrade pattern is visible in a CIM if you know what to look for. I have seen an investor walk away from a target for exactly this reason, because the movement between high-value and low-value subscription tiers was heading the wrong way while headline retention looked healthy.
For anything that has been assembled, ask what is actually running. Sellers naturally present acquired portfolios as integrated platforms. A few specific questions usually establish what has genuinely been merged. I have seen a case where the honest answer, five years after the acquisitions, amounted to single sign-on and very little else. The investor walked, because what was on offer was a zoo of unintegrated companies rather than a scalable fleet management platform. Ask how many production platforms are running today, how many engineering teams are maintaining them, and what the dual-running cost line looks like.
Check the installed base age profile against the next network shutdown. When a mobile network generation is switched off, every device that depends on it has to be replaced, on a timetable nobody in the business chose. A base that is due one shortly is carrying a cost the model may not show, and it interacts badly with any migration already in progress.
Where I Come Out
The discount is real and it is blunt. Fleet management assets are being priced as software at a moment when software is being marked down for AI exposure, and the exposure in this industry is genuinely lower than in horizontal SaaS for reasons that are physical, contractual and regulatory rather than clever. At the same time, part of the correction is deserved, some of it predates AI entirely, and the buy-and-build thesis that drove much of the last decade’s activity has real problems that have nothing to do with the current news cycle.
That combination is exactly the condition in which somebody overpays and somebody else gets a bargain, and the difference between the two is not visible in the headline metrics.
I have my view on which side of this the average asset sits. I am more interested in yours, particularly if you are looking at something right now.
