Summary
Friedberg’s prediction was that as per‑mile/dynamic auto insurance gained adoption, low‑mileage/good drivers would move to these products, forcing incumbent insurers to raise rates on the worse‑risk customers left behind, which in turn would create a “huge market moat” for per‑mile insurers over the following years. By late 2025, the evidence points the other way: while telematics/usage‑based insurance (UBI) has grown, large incumbents dominate it, specialized per‑mile startups have not built a moat, and industry rate hikes are driven by other factors (inflation, claims costs, weather) rather than documented adverse selection from per‑mile migration.

Key points against the prediction

1. Per‑mile specialists did not build a competitive moat; they struggled instead.

   • Metromile, one of the original pay‑per‑mile pioneers and the likely reference for “our service,” went public via SPAC in 2021 and then collapsed in value; Lemonade acquired it in July 2022 for under $145M in stock, versus a prior valuation over $1B, and promptly laid off ~20% of staff. (techcrunch.com)
   • After the acquisition, Metromile’s written premiums declined: by mid‑2023 it had about $49M in written premiums, ~15% less than a year earlier, with decreases in every state, showing contraction rather than moat‑like growth. (coverager.com)
   • More broadly, analyses of the U.S. usage‑based/telematics market list major incumbents — Progressive, Allstate, State Farm, Liberty Mutual — as the primary players by share; specialized insurtechs like Metromile and Root are small by comparison. (futuremarketinsights.com)

2. Dynamic/telematics pricing is now led by incumbents, not by a separate per‑mile niche.

   • Market reports on telematics‑based and usage‑based auto insurance show rapid growth but emphasize that the segment is dominated by large traditional insurers (Progressive, Allstate, State Farm, Liberty Mutual, etc.), who have rolled out their own dynamic‑pricing and pay‑per‑use products at scale. (futuremarketinsights.com)
   • Examples: State Farm’s Drive Safe & Save, Progressive’s Snapshot, Allstate’s Drivewise, Nationwide’s SmartRide and SmartMiles (explicitly pay‑per‑mile for low‑mileage drivers) are all usage‑based products offered by incumbents. (statefarm.com)
   • So instead of per‑mile specialists pulling good risks away from incumbents, incumbents themselves are offering the same or similar telematics/low‑mileage products, blunting any moat for standalone per‑mile insurers.

3. Adoption of telematics/UBI is meaningful but still a minority of the market.

   • A 2025 survey of U.S. drivers finds only about 12% currently enrolled in an insurer telematics program; 88% are not. (autoinsurance.com)
   • A Maryland regulator report shows similar figures: roughly 13.2% of auto policies in that state were in a telematics program in 2023 (up from 9.5% in 2021), indicating steady but not overwhelming penetration. (insurancejournal.com)
   • Global and U.S. UBI/pay‑per‑mile markets are growing quickly in dollars but remain a relatively small slice of the overall auto insurance market, which is on the order of hundreds of billions annually in the U.S. alone. Estimates put pay‑per‑mile globally around $8.2B in 2024, projected to ~$32.7B by 2033 — material, but far from dominating auto insurance. (marketintelo.com)
   • With such partial adoption, the kind of severe adverse‑selection spiral Friedberg described (good drivers en masse leaving incumbents) has not materialized at industry scale.

4. Auto insurance rate hikes since 2020 are explained by inflation and loss costs, not by losing good risks to per‑mile.

   • From 2020 to 2024, U.S. motor vehicle insurance costs rose about 54%, but analysis attributes this primarily to inflation, sharply higher vehicle repair and replacement costs, medical inflation, higher claim severity, and climate‑driven catastrophe losses, plus investment‑income pressures — not to an exodus of low‑risk drivers to pay‑per‑mile insurers. (usafacts.org)
   • Major insurers’ own explanations for recent rate hikes align with those factors (repair labor and parts, more severe accidents, weather, litigation) rather than adverse selection from telematics/pay‑per‑mile programs. (thezebra.com)
   • In fact, incumbents often enroll their better risks into their own telematics programs and give them discounts, keeping many good drivers within their portfolios rather than ceding them to niche per‑mile startups. (autoinsurance.com)

5. Some telematics‑first startups are improving, but still without a clear moat over incumbents.

   • Root, a telematics‑heavy personal‑auto insurtech, has recently posted its first profitable year and strong stock performance, but it remains tiny relative to Progressive or Allstate in policies‑in‑force and premiums. (investors.com)
   • At the same time, incumbents like Geico, which initially lagged in telematics, have been making “rapid strides” and are now described by Berkshire Hathaway leadership as competitive with peers on telematics capabilities — again undercutting the notion that startups enjoy a durable moat. (reuters.com)

Conclusion
Over almost five years since the November 2020 prediction, we do see:

• More usage‑based and pay‑per‑mile options in the market;
• Low‑mileage/safe drivers sometimes benefiting from such programs; and
• Significant overall auto premium inflation.

However, the specific mechanism Friedberg forecast — low‑risk drivers flocking to per‑mile specialists, forcing incumbents to raise rates on a deteriorating pool and thereby creating a “huge market moat” for per‑mile insurers — has not occurred in observable data. Instead, telematics has been widely adopted by incumbents themselves, per‑mile specialists like Metromile have shrunk or been absorbed, and rate hikes are driven by cost and loss trends rather than documented adverse selection from per‑mile migration.

On balance, the prediction is wrong as of late 2025.