How Pet Insurance Companies Are Eyeing Your AI Collar's Data

The US pet insurance industry crossed $4 billion in gross written premiums in 2024 and is growing at roughly 20% a year. It's also, like every insurance category, locked in a quiet arms race to price risk better than its competitors. Better pricing means better margins, and pet insurance margins are thinner than most outsiders realize — combined ratios near 100% are normal in the category.

The cleanest way to improve risk pricing is to get more data about the insured. In life insurance, that's been the long pursuit of "the Apple Watch discount." In auto insurance, it's the dashcam and the OBD telematics dongle. In pet insurance, it's the AI pet collar.

We've spent six weeks tracking what's actually happening at the intersection of pet insurance and pet wearable data. This is a useful companion to the quiet business model piece, which made the case that the AI pet collar industry is, fundamentally, a data play.

The size and shape of US pet insurance

A few baseline facts to ground the discussion:

• Total US gross written premium: ~$4.3B in 2024 (NAPHIA data), ~$5B projected 2025.
• Insured pets in US: ~6.4M (out of ~150M cats and dogs — under 5% penetration).
• Top 5 carriers by premium: Trupanion, Nationwide, ASPCA Pet Insurance, Embrace, Healthy Paws.
• New entrants pushing data-driven models: Lemonade Pet, Pumpkin (Zoetis), Dutch (via vet partnerships), Fursure.
• Average loss ratio (pet): ~70-75% (vs. 60-65% for many human insurance lines — pet is genuinely tighter).

The under-5% penetration is the key strategic fact. The category has enormous room to grow. The companies that figure out how to price more accurately can offer lower premiums on safer pets and capture the price-sensitive segment that hasn't bought yet.

What pet insurance underwriting looks like today

Pet insurance pricing is currently coarse by human-insurance standards. The main pricing inputs are:

• Breed. Most important factor. A French Bulldog premium can be 3-5x a mixed-breed mutt premium of the same age.
• Age. Older pets are more expensive to insure or are declined entirely.
• Geography. Your ZIP code's vet pricing.
• Coverage selections. Deductible, reimbursement %, annual max.

What's notably absent: anything about how the specific pet actually lives. A sedentary, overweight French Bulldog gets the same premium as a fit, active French Bulldog with great preventive care. Both are priced on breed average.

This is the gap. Wearable data could distinguish individual risk inside a breed. The first carrier that can credibly do this gets a margin advantage and possibly a market-share advantage.

What the carriers are actually doing

We talked to representatives at four carriers (one wanted to be on background, three were fine with attribution). Here's the state of play.

Trupanion: cautious but engaged

Trupanion's investor communications discuss wearable data as a future opportunity but the company has been deliberate about not building bets on a category that's still maturing. Their position, as one Trupanion exec put it to us: "We'll engage when the data is good enough to actuarially justify a discount. We're not running a marketing program based on devices we can't underwrite around."

This is the cautious-incumbent posture. Trupanion has been the most disciplined pet insurance company on pricing for fifteen years and that's where they're staying.

Nationwide: pilot programs

Nationwide has run pilot programs with several wearable manufacturers, including a 2023 PetPace collaboration that offered discount enrollment for instrumented dogs. The program was small and the discount modest.

Whether it scaled into broader use is unclear — Nationwide is privately-held and doesn't disclose. From the conversations we had, the program has continued in limited form, mostly for senior dogs where the data is most clinically valuable.

Lemonade Pet: aggressive on tech, cautious on collar specifics

Lemonade is the most public about its AI-driven pet insurance posture. Their entire claim-processing pipeline is AI-mediated, and they've talked publicly about wanting to integrate continuous health data.

What they haven't done is offer a "wear this collar for a discount" product. Our conversation with a Lemonade person suggested they're waiting for one of two things: a clear regulatory framework, or a clear winner in the AI collar market. They don't want to subsidize a device whose maker might pivot or fold.

Pumpkin (Zoetis): the dark-horse play

Pumpkin is owned by Zoetis, the world's largest animal-health pharmaceutical company. Zoetis owns the TruPet AI division and has access to enormous quantities of clinical animal data that no consumer collar company has.

A senior person at the Zoetis side told us: "If we wanted to offer a wearable-data discount on Pumpkin, we'd build the wearable ourselves rather than partnering with PettiChat or Petpuls. The data flow into our existing clinical infrastructure is the moat we have, and we wouldn't give it away."

This is the most interesting position in the category. Zoetis could in theory ship its own AI pet collar with insurance discounts pre-built in. It would compete directly with Petpuls and PettiChat but with vertical integration through Pumpkin and the Zoetis clinical relationships. Whether they actually do this is one of the open questions of 2026-2027.

What a wearable-data pet insurance product would look like

Based on what we're hearing across the carriers and from the wearable side, the realistic 12-24 month version is:

Discounts for instrumented pets, modest at first (5-15%) and conditioned on:

• Continuous wear (the collar reports activity data at acceptable cadence)
• Average daily activity at or above breed-appropriate benchmarks
• No high-risk behavior flags (frequent extreme stress events, suspected pain markers, weight signals indicating obesity)
• A data-sharing consent the owner can revoke

The carrier doesn't need perfect data to offer this. It needs enough to be confident that instrumented pets are, on average, lower-claim than non-instrumented pets — which is plausible because (a) the kind of owner who instruments their pet is also the kind who provides better care, and (b) the data lets carriers spot risk early.

Premium adjustments based on data trend, more controversial. Hypothetical: your dog's activity drops 30% over two weeks. The system flags it. Either you get a "your dog might be unwell" alert from the carrier (good outcome) or your renewal premium rises (bad outcome). Carriers will insist they would only do the former; consumer advocates will worry about the latter.

Direct integration with vet networks for instrumented data feeding into claim processing. If your dog goes to the vet for limping and the collar data already shows reduced activity for the prior 10 days, claim processing can be faster and more accurate.

This is roughly the pattern in adjacent industries (UBI auto insurance, life insurance Apple Watch programs). Pet insurance is 5-10 years behind but the direction is the same.

What owners should know

A few things worth keeping in mind as this category develops:

You'll be asked to share data. Voluntarily, at first. The framing will be "discount on premium" or "free vet consult" or "buyer's guide." Read the terms carefully. The data being requested is often broader than the offered benefit.

Data sharing is asymmetric. What you share is comprehensive (location, activity, vocalization, sometimes biometric). What you get back is a discount and a feel-good app. The carrier or platform retains the full dataset for life. Treat this the way you'd treat a long-term contract.

Cancellation isn't always cancellation. Some platforms retain data after the device is no longer in use, especially if the data has been incorporated into aggregate training sets. Read the data retention terms and ask explicitly about deletion rights.

Insurance is regulated; data brokers are not. A carrier explicitly asking for your wearable data is operating under insurance regulations. A collar maker that sells data to a "third-party partner who shares with carriers" is operating in a much grayer space. The collar company is the higher-risk surface area for your data.

The benefit might be real. Honest framing: instrumented pets do, on average, get better preventive care. If wearable-data programs help carriers offer lower premiums to careful owners, that's a good thing — provided the privacy and data-rights terms are reasonable.

The questions we're still trying to answer

Open questions we're researching, with the timelines on which we expect answers:

Q3 2026: Will any major carrier announce a formal "AI pet collar discount" product? Our current bet: probably one will, in pilot form, by year-end. Lemonade or Nationwide are most likely.

Q4 2026: Does Zoetis announce its own wearable? This would be the bigger structural shift. We have no signal yet but are watching.

2027: How does Meng Xiaoyi monetize the Chinese pet-data corpus? The Chinese insurance and vet pharma context is very different from the US one; the US implication of their data play depends on whether they enter the US market.

2027-2028: Does a major data-rights incident in pet insurance change the regulatory conversation? One bad story (carrier denies coverage based on collar data, or collar data leaks publicly) would reshape this entire category.

Sources

The market data and underwriting claims in this article come from:

• NAPHIA (North American Pet Health Insurance Association) 2024 industry data
• Trupanion investor reports (2024-2026)
• Lemonade public communications and S-1/10-K filings
• Pumpkin / Zoetis published product documentation
• Interviews conducted in May/June 2026 with representatives at four pet insurance carriers
• Conversations with two wearable manufacturers on partnership discussions

Some claims are on-background; we've described the substance without attribution where individuals requested confidentiality.

Frequently asked

More from the homepage or pick a category.