How many rides does a robotaxi actually give per day?

Ask what a robotaxi earns and every answer decomposes the same way: rides per day, times revenue per ride. The second factor at least has published fares. The first is the ghost in every spreadsheet — assumed, never sourced. So let's source it.

The division nobody shows you

Two snapshots, roughly a year apart, both built entirely from disclosed figures.

April 2025. Waymo announced it was serving 250,000 paid rides per week (April 24, 2025, up from 200,000 in February). Eleven days later, its own blog described the fleet as “over 1,500” vehicles (May 5, 2025). Divide: 250,000 ÷ 7 ≈ 35,700 rides a day; 35,700 ÷ 1,500 ≈ 24 paid rides per car per day.

March 2026. Waymo crossed 500,000 paid rides per week across 10 cities (Sherwood, March 30, 2026). Three days earlier, TechCrunch put the fleet at roughly 3,067 robotaxis as of December 2025, noting the fleet had held roughly steady while trips grew. Divide: 500,000 ÷ 7 ≈ 71,400 rides a day; 71,400 ÷ 3,067 ≈ 23 rides per car per day.

The alternative denominator matters. A May 6, 2026 recall filing covered 3,791 Waymo vehicles, up from 1,212 a year earlier (FuelArc, May 16, 2026) — but recall populations include vehicles that are not necessarily in active ride service, so dividing by 3,791 gives you a floor of ~19, not a better estimate.

That steadiness is the genuinely remarkable finding here. Doubling a fleet usually dilutes per-unit utilization; Waymo doubled and held ~23–24.

About 150 miles a day

Same method, different disclosed numerator. Waymo reported 4 million-plus fully autonomous miles per week (Waymo blog, June 24, 2026). Against a fleet of ~3,800: 4,000,000 ÷ 7 ≈ 571,000 miles a day; 571,000 ÷ 3,800 ≈ ~150 miles per car per day.

For growth context: 50,000-plus weekly rides in May–June 2024, 100,000 by August 20, 2024 (TechCrunch), 250,000 by April 2025, 500,000 by March 2026. Ten-x in under two years — while rides per car stayed flat.

The number nobody puts in the spreadsheet: empty miles

Here is where the honest math gets uncomfortable. A peer-reviewed study in Findings (May 19, 2026, with an MIT Transit Lab author) analyzed Waymo's first 1,000 days of commercial service in California: 86.3 million vehicle miles traveled, 13.79 million trips — and only about 54% of those miles carried a passenger. The share started near 36% in the early months and plateaued around 55–57%. The one clear improvement: the average empty leg driven to a pickup shrank from 5.1 miles to 2.8.

Run the derived division: 86.3M miles × ~54% ≈ 46.6M passenger miles; 46.6M ÷ 13.79M trips ≈ ~3.4 passenger-miles per average trip. That is a derived figure, California-only, and heavily weighted toward short San Francisco trips — label it accordingly, because we're about to use it.

Before anyone gasps at 46% empty: that is roughly the same as human ride-hail. The same Findings paper cites prior research putting Uber and Lyft deadheading above 40% of vehicle miles. And Cramer & Krueger's NBER study (w22083, March 2016) measured UberX drivers carrying a passenger about half their working time (46.8% in Boston to 54.9% in San Francisco) and about 61% of their miles (64.2% in Los Angeles, versus 40.7% for taxis there). The fair conclusion: robotaxis haven't beaten the empty-miles problem. They've matched it.

Demand isn't flat, and neither is the map

Averages hide shape. Ride-hail demand piles into Friday and Saturday nights and weekday rush hours — Chicago's regional planning agency documented exactly that pattern in its trip-level TNC data (CMAP, May 4, 2019). And demand clusters in space as much as time: airports alone account for 15% of Uber's global mobility gross bookings (Uber Engineering, August 19, 2025). That is why robotaxi geofences hug airports — a dynamic we walked through for Orlando specifically. A car that averages 23 rides a day is not giving one ride an hour; it is slammed at peaks and idle at 4 a.m.

The Tesla column of this table is blank

Everything above is Waymo, because Waymo is the only operator disclosing enough to divide. Tesla publishes no ride counts, no official fleet size, and no utilization figures for its robotaxi service — even as the service scales quickly. The dated record: seven live metros per Tesla's Q2 2026 report (July 22, 2026), six of them running unsupervised; roughly 2.5 million cumulative paid miles by that report, against Waymo's 4 million per week; and community tracking — unofficial, since Tesla confirms no count — that put about 21 active cars on the road in late July and about 85 by August 26, 2026. That same day Tesla widened operating hours to 6 a.m.–10 p.m. across every market, said the unsupervised fleet had grown “a lot bigger,” and shipped smarter vehicle distribution aimed at wait times (Not a Tesla App, August 26, 2026). More cars on longer days move both sides of the division — but until Tesla discloses a numerator or a denominator, nobody outside the company can compute rides per car, and Tesla isn't saying.

What 20–24 rides a day means for owner math

Everything in this section is a projection, assembled from the derived averages above — not a disclosed figure from any operator. Take a deliberately conservative 22 rides a day and the derived ~3.4-mile average trip. At the fares announced for Orlando's robotaxi launch — from $3.25 base plus roughly $1 per mile — an illustrative ride runs about $3.25 + $3.40 ≈ $6.65, and 22 of them ≈ ~$146 a day gross, call it ~$1,000 a week. Projection, illustration, not a promise.

One cross-check that the pieces cohere: 22 trips × 3.4 miles ≈ 75 passenger-miles; at a ~54% passenger-mile share that implies ~139 total miles driven — right on top of the ~150 miles per day we derived independently from Waymo's mileage disclosure. The numbers agree with each other.

Also remember what “gross” ignores: platform fees, depot operations, commercial insurance, charging, cleaning, and depreciation all come out before an owner sees anything — the full stack is in our honest Cybercab P&L, and you can set every assumption yourself in the earnings calculator. A short-trip, SF-weighted 3.4-mile average may also simply be wrong for a spread-out, airport-heavy market like Orlando — longer trips would mean fewer rides at higher revenue each.

What nobody publishes, so you don't over-trust the math

• Per-vehicle utilization and revenue per ride — no operator discloses either. Everything above is division, not disclosure.
• Average Waymo ride length — there is no official figure. Our ~3.4 miles is derived, California-only, and SF-heavy.
• Tesla's numbers — all of them. Rides, fleet, utilization: unpublished.
• Whether 24/7 availability can flatten the peak curve — a robotaxi doesn't sleep, but riders do. Unproven either way.
• Waymo's next milestone — the widely-discussed ambition of a million rides a week by the end of 2026 is a target, not a result. We'll do this division again when there's a new numerator.

Frequently asked questions

How many rides does a Waymo do per day?

Roughly 23–24 paid rides per vehicle per day, derived from disclosed totals — Waymo publishes no per-vehicle figure. April 2025: 250,000 weekly rides ÷ ~1,500 cars ≈ 24. March 2026: 500,000 weekly rides ÷ ~3,067 cars ≈ 23 — or ≈19 against the 3,791-vehicle May 2026 recall population, which includes cars not necessarily in active service.

How much money can a robotaxi make in a day?

No operator publishes revenue per vehicle, so any dollar figure is a projection. As a clearly-labeled illustration: ~22 rides at the derived ~3.4-mile average trip, priced at Orlando's announced fares (from $3.25 plus ~$1 per mile), is roughly $145–150 a day gross — before platform fees, depot costs, insurance, charging, and depreciation. Built from derived averages, not disclosed figures; run your own assumptions instead of trusting ours.

How many Waymo cars are there in 2026?

Roughly 3,000–3,800 depending on the count: ~3,067 robotaxis as of December 2025 (TechCrunch, March 27, 2026), and 3,791 vehicles in a May 6, 2026 recall filing — a ceiling, since recall populations include vehicles not necessarily serving riders. The fleet was “over 1,500” in May 2025, with 2,000-plus more I-PACEs planned through 2026.

Do robotaxis drive around empty between rides?

Yes — roughly 46% of Waymo's California miles carried no passenger across its first 1,000 commercial days (86.3M miles, 13.79M trips; Findings, May 19, 2026). The passenger-mile share climbed from ~36% to a 55–57% plateau, and the average empty pickup leg shrank from 5.1 to 2.8 miles. That roughly matches human ride-hail: Uber/Lyft deadheading exceeds 40% of miles, and NBER data from 2016 put UberX passengers in the car about half of drivers' working time and ~61% of miles. Matched, not beaten.