Reading · Jonathan Levitt

Disclosure: TrainingPeaks is a partner of mine, and I have friends at both organizations. That gives me context many won’t have, and a bias worth naming up front. I’ve tried to write the honest version anyway.

Peaksware, the parent company of TrainingPeaks and TrainHeroic, sits in Louisville, about fifteen minutes from my front door. So when Garmin announced today that it had acquired both platforms, the news landed close to home, pretty literally.

Most of the coverage is reading it the obvious way. Garmin consolidates the data, Garmin users get one ecosystem, everything syncs in one place. That’s real, and it’s also the least interesting thing about the deal.

You can find the scale in about ten seconds. Garmin runs a business north of $7 billion in revenue, with more than 23,000 employees across 100-plus offices and roughly 40 million people on Garmin Connect. I’ve spent a decade in meetings with their team and they’ve mostly been focused on selling hardware. They release 80+ new products per year. The questions to ask are what Garmin gets that it couldn’t build, and who benefits first.

One narrative is worth tossing out up front. This is not a hardware company that has fully ignored software and suddenly discovered it. Garmin launched Connect+ in March 2025, its first paid subscription tier, built around the AI insights it calls Active Intelligence, and it has told investors that new advanced features will live behind that paywall. So the real story is acceleration. Garmin had already started building the software and subscription layer, and TrainingPeaks lets it skip the decade of trust the coaching world would otherwise make it earn from scratch, a brand as big as Garmin might earn that trust faster than most, but software has never been its strength, which is a big part of why buying the coaching standard beats building it.

And this is act two of a familiar playbook. Garmin acquired Firstbeat Analytics, the Finnish company whose algorithms turn raw heartbeats into Training Status, Body Battery, and VO2 max, back in 2020. That deal gave Garmin the engine behind the metrics on your watch. TrainingPeaks gives it the fuel: structured, human-graded training data that can make those metrics sharper and, before long, train a model on how real coaching is actually done.

TrainingPeaks’ primary customer has never been the athlete. It is the coach. Athletes come and go. The coach is the one paying, building a business, and staying for years.

What the acquisition hands those coaches is distribution. A coaching business on TrainingPeaks now sits one step away from an ecosystem of roughly 40 million Garmin Connect users, and it can reach them no matter what the athlete straps to their wrist. The coaches who lean into that, rather than brace against it, are the ones positioned to grow, and they can do it with clients on any brand of watch.

This isn’t only my read. The people I've talked to who are close to the deal land in the same place. My bet in endurance has always been that the durable value lives in the small, high-trust nodes and the infrastructure underneath them, well ahead of raw reach or another apparel logo. The coach is the highest-trust node in the sport. Garmin just bought the layer those relationships run on.

Corrine Malcolm, a trail running coach and athlete who has been on TrainingPeaks for a decade, asked me this today, knowing I have a few friends and contacts at both organizations: “Have you heard anything on whether this changes anything for coaches using a platform that’s historically been hardware agnostic? Otherwise Final Surge is going to get a nice bump.” She figures that concern will be commonly felt, and I think she’s right that it’s the first thing most coaches will reach for.

Here is why I don’t think it comes to pass. The economics line up with staying open. TrainingPeaks has spent 26 years integrating with nearly every device on the market, and there is a hard financial reason to expect that to continue.

The logic is simple. The coach marketplace is only worth what Garmin paid if it stays large, and it only stays large if a coach can train a client on a Coros, a Wahoo, a Polar, or an Apple Watch without friction. Wall it off to Garmin devices and you shrink the thing you just bought. Open is the profitable choice here, which is a better guarantee than goodwill.

The item to watch is subtler than the open-versus-walled question. It is whether Garmin-device athletes slowly get features the rest do not, the way Connect+ already gates the good stuff. Alignment today does not guarantee alignment once someone inside the building models the subscription upside.

Not every coach is going to read this as good news, and it would be naive to pretend otherwise.

Troy Meadows, a coach who runs his business on TrainingPeaks, is taking a wait-and-see line, and his caution is about integration rather than devices. His question is whether TrainingPeaks stays “siloed and allowed to operate on their own” with Garmin’s operational and legal muscle folded in, or whether “Garmin is gonna gut the team in a year and run the show from corporate.” He has watched this play out before: what an acquirer promises at the table to get a deal signed, he notes, is usually not how things shake out for the acquired team a year or two later.

That is the real uncertainty, more than the device question. Deal-table language is cheap, and the truth tends to show up in year two, in whether Garmin treats TrainingPeaks as an asset to protect or a team to absorb. Depending on a single large owner is a genuine risk, and the honest version of this story holds both outcomes at once: more distribution and more dependence, arriving together.

The early signals point the gentler way. Garmin is keeping TrainingPeaks' president, Lee Gerakos, in charge of the platform, and co-founder Dirk Friel is staying on as a strategic adviser. History matches this trend too: the last time Garmin made an acquisition like this, it left the target largely intact, and Firstbeat still runs out of Finland, where Garmin later opened a dedicated lab for it. Both point toward the optimistic case. Keeping leadership in place on announcement day is standard, and a 120-person platform with a coaching community attached is a harder thing to leave alone than an analytics team. For a variety of reasons, I hope the optimistic case is reality; I have a lot of friends at TrainingPeaks and they’ve done a fair amount of business with the Long Run Labs Network too.

Garmin is not the only one consolidating. Across 2025, Strava bought its way into the same layer, acquiring the running app Runna in April and the cycling app The Breakaway in May, then bolting AI training plans onto their app, which has more than 150 million users.

The two plays point at different customers, which is what makes the contrast worth drawing. Strava’s bet is that a good-enough algorithm can stand in for a coach. TrainingPeaks is making the opposite bet, that the human coach is the product and the software’s job is to make them stronger. Same category, two very different theories of where the value lives.

There’s a data dimension underneath that split, and it’s the one that matters most. Garmin and Strava both hold activity files at a scale nobody else can match, but activity only tells you what an athlete did, never what they were told to do or by whom. TrainingPeaks has the missing half. For 26 years, across every brand of watch and head unit, it has logged what a qualified coach prescribed, whether the athlete actually completed the session, and what they did in the weeks that followed. As the5krunner argued in a sharp breakdown of the deal, that gap is the whole game: activity data alone can teach a model what athletes did, while the TrainingPeaks archive can teach it what a coach chose and whether the choice worked. It is also the raw material for fixing the training-status analytics Garmin users love to hate, the ones that hand you an “Unproductive” verdict with no explanation. Garmin could buy that capability once, from one company, and now it has. The real prize is the data set, not the model.

For everyone else, the ground just shifted. Hardware-and-software rivals like Coros and Wahoo now compete against a Garmin that owns the sport’s coaching standard. And the platform-neutral independents, the ones like TrainerRoad, Intervals.icu, and Final Surge, get more interesting, not less. Corrine’s instinct points right at it: if this changes for coaches, the neutral option is suddenly worth more, precisely because the giants are buying up everything around it.

Here is the flip side of the distribution upside, and it’s the part coaches should sit with. That archive, the most valuable thing Garmin just bought, was built session by session by the coaches themselves. They are TrainingPeaks’ primary customer and, at the same time, the people generating the exact data Garmin would need to build coaching software of its own.

So an AI coach of some kind is a safe bet, trained on the accumulated judgment of human ones. The lazy version of this story ends with the machine replacing the humans, and I wouldn’t bet on that story. What an AI coach actually does is split the market. Athletes who were never going to hire a person, the self-coached and the never-coached, get a real option for the first time, which grows the top of the funnel more than it threatens anyone. And perhaps it’s even a pipeline from AI coach to human coach, for those who get a taste of coaching for the first time and want more. Athletes who want a human, for the accountability, the trust, and the read on a life that never shows up in the data, still choose one. The model reaches the unserved and the low end. The relationship at the core of good coaching it mostly leaves alone.

That is better news for coaches than the doom and gloom version, but it is not all good news. The tier most exposed is the entry-level human coach, whose offer sits closest to what a decent algorithm can now do cheaply, and the thing training that algorithm is the work those very coaches uploaded. How (and if) Garmin prices and positions an AI tier, whether it sits beneath the human coaches or leans on them, is what decides whether this grows the pie or squeezes the bottom of it.

There’s also a consent question hiding in here. That data was created inside coaching relationships, not as training fodder for a model, and whether anyone asks the coaches and athletes who made it is the sort of thing that has become a real fight in other creative fields. Endurance won’t be immune.

TrainingPeaks’ leadership would put all of this more warmly. In announcing the deal, co-founder Dirk Friel called it a doubling down on the company’s founding mission of empowering human coaches. I think that can be true and still incomplete, since the same data that helps a human coach can also train the tier that undercuts the cheapest of them. So if you coach for a living, the question isn’t whether the machine takes your job. It’s whether you’re doing the part of it a machine can’t. It’s the art side of the art and science of coaching.

Step back and look at what Garmin is assembling. Passive capture around the clock through Cirqa, the screen-free band it just launched this week to compete with Whoop and others. Structured prescription and analysis through TrainingPeaks and TrainHeroic. An AI insight layer through Connect+. Measure, prescribe, analyze, adapt. It is close to a closed loop, and the piece still missing is the why underneath the physiology. And I can’t wait for you to read the next few words and have an a-ha moment too.

That piece is blood. Here is the version I’ve wanted to see built for years. Imagine the platform watches your data over weeks, not days. Your sleep quality has been sliding, or your easy runs keep drifting harder with your heart rate creeping up at paces that used to feel simple, or you come apart every time you go train at altitude. Instead of leaving you to guess, the system says this might be worth a blood test. Then it lets you order one in a click and, the harder part, tells you what the results mean for your training.

The signals map onto real markers. Sliding sleep can trace back to something nutritional like low vitamin D or magnesium, or to load, stress, and under-fueling, and a panel that looks at cortisol, lipids, and hormones including testosterone can point toward which. Easy efforts turning heavy is a classic under-fueling or iron story. Coming apart at altitude sends you straight to iron and ferritin, the oxygen-carrying markers. None of this replaces a doctor, but as a directional nudge inside a training platform it is useful.

I pitched a version of this to TrainingPeaks years ago. The opportunity wasn’t there yet. It is now, or it’s close. And the part people underrate is what it does for coaches: a blood-driven prompt gives a coach one more reason their athletes live on the platform, and one more lever in the relationship. The ecosystem unlock runs both directions, more athletes discovering coaches, and more reasons for everyone to stay.

This isn’t a wild guess. WHOOP and Oura have already bolted consumer blood testing onto their ecosystems, both under $200, and Ultrahuman is doing the same. Garmin is the conspicuous holdout among the big wearable players. I wrote a whole piece on where this blood-and-wearables convergence is heading, in The New Blood Era; the short version is that testing is commoditizing and the value is moving to interpretation, which is exactly the layer Garmin would be buying.

So when I look at Garmin’s next move, I don’t see another sensor. The capture side is commoditizing fast, with Fitbit Air, Polar Loop, and Amazfit all crowding the screen-free category Cirqa just entered. Interpretation is where the value sits now, and it is still hard to do well. If Garmin keeps following the loop it is building, the blood-testing and interpretation layer is the category it reaches for next. I won’t name targets, and I don’t have to. Anyone who has looked at this space knows the short list.

So no. This was never about consolidating your data into one tidy place, though I do appreciate when that happens. Garmin bought your coach’s distribution, a 26-year head start on trust it could not have manufactured, and the richest labeled training set in endurance sport. Whether that turns into something coaches love or something they tolerate depends on choices Garmin has not made public yet.

What do you think? I’m curious to hear what you agree or disagree with!

Garmin didn’t buy a training app. It bought a coaching network and a mountain of data.

Comments