Did Tesla Just Solve Robotaxi Economics? NHTSA Rules, Dynamic Pricing, and the Subscription Play

Tesla's robotaxi business model is being assembled piece by piece, and three pieces clicked in the same week: the rules, the pricing, and the subscriptions. In the Brighter with Herbert conversation, chartered financial analyst Cern Basher of Brilliant Advice stitches those threads into one economics story. The claim is bold, but the logic is simple: if a federal framework asks 'is this a safe driver?' instead of 'does it have mirrors?', if price is set for the whole network rather than one ride, and if billing moves to once a month, a taxi business stops being a taxi business and becomes a subscription business.

The regulatory headline is NHTSA Administrator Jonathan Morrison's reset. Instead of checking prescriptive design boxes, the agency wants to test whether the vehicle drives safely. Morrison's line lands well: a scalpel, not a hatchet. Today's Federal Motor Vehicle Safety Standards were written around a human driver, so they tether future designs to past assumptions. Updating them around performance requirements, without lowering the safety bar, would let creative designs through while holding them to outcome tests.

Rules: Performance Over Prescription

The first move covers five standards and each is instructive. FMVSS 101 and 102 demand a transmission position display (P-R-N-D) that means nothing in a driverless Cybercab. 103 and 104 require windshield defrosting and wiping hardware that matters only if the driver sees through glass, not if the driving system sees through sensors. 110 calls for a tire and loading placard on the 'driver side B-pillar,' which in a vehicle with no driver side becomes 'left side.' 135 says service brakes shall be activated by a foot control, while a pedal-free vehicle brakes through actuators and software. Keeping stopping-distance performance while dropping the foot-pedal prescription is the core of the June 2026 FMVSS 135 proposal.

More is coming after the first five. Mirrors, inside rearview, rear visibility, turn-signal cancellation by steering rotation, and stability-control telltales sit on the next list. To avoid writing these alone, NHTSA is standing up the ASCEND consortium with standards body SAE to accelerate performance tests for automated driving systems and to develop a national competency benchmark. The shift is structural: not a one-off patch, but a move from design checklists to driving performance.

The timing is the tension, and Tesla is not waiting. In the United States makers self-certify compliance and NHTSA audits after the fact. Amazon's Zoox took the permission route and in July 2026 received a temporary Part 555 exemption covering eight standards including the foot-brake and mirror rules before offering commercial service. Tesla took the certification route, declared the Cybercab compliant with every applicable FMVSS, and started commercial rides in Austin on September 3. NHTSA opened an Audit Query on September 4 and escalated to a 21-question Special Order on September 10 demanding a sworn explanation.

The Special Order is blunt: answer by September 30, fully and truthfully, or face penalties up to $139 million. Addressed to Tesla's litigation and regulatory counsel, it cites the statute that no person may issue a certification if, with reasonable care, they have reason to know it is false or misleading in a material respect. The agency expects Tesla to argue that many rules simply do not apply to a vehicle with no human driver, but reminds the company that until rules are formally rewritten, 'inapplicable' is not a unilateral checkbox.

The 21 questions read like a regulatory audit outline. Scaling plans and locations for subject vehicles, how compliance was established and evidenced, whether temporary human controls were used to certify and then removed, and how that squares with the 'make inoperative' prohibition. Then the standards walk-through: 101 controls and telltales, 102 shift display, 108 turn signals that cancel with steering-wheel rotation, 111 mirrors and rear-camera images, 126 stability telltales, and the sharpest one, 135, where the plain text 'activated by means of a foot control' meets a car with no foot control at all.

Tesla's expected answer, rehearsed in the video with help from Grok, has three layers. First, rules written for a human driver are not applicable where there is no human driver. Second, performance is met or exceeded by other means, with braking distance and system redundancy handled by actuators rather than a pedal. Third, practical bridges for operations: a touchscreen joystick to inch the vehicle forward or back and training for first responders to use it. NHTSA has proposed to rewrite 135 for ADS-equipped vehicles without manual controls, but a proposal is not law until finalized, and the agency is enforcing the current text in the meantime.

Pricing: Dynamic, Not Just Surge

On pricing the video does the most clarification. Surge pricing is a subset. It raises price when supply is scarce. Dynamic pricing is the superset that sets price for the value of the trip to the entire network. Cern Basher's point is that Tesla will not only chase the surge, it will price to move cars to where the next demand will be.

The stadium example makes the mechanism tangible. The end of a game is predictable, traffic patterns around stadiums repeat, and AI can preposition vehicles hours early. While repositioning, those vehicles do not have to run empty; they can carry a rider in the opposite direction for a low fare and earn revenue on the repositioning leg. An idle Cybercab earns nothing, a moving one always has a chance to earn per hour. The optimization is as much about where to put cars as what to charge for them.

From there the idea turns into network pricing, a live market like a stock exchange. The price reflects how many vehicles are nearby, how many requests are incoming, what traffic will be in 20 to 40 minutes, where the trip will strand the car for the next ride, state of charge, and whether the car needs cleaning. Tesla controls supply directly, unlike Uber which nudges human drivers with incentives that drivers have learned to game by staying out of a surge zone and flooding it once the price spikes. With a driverless fleet the game disappears, and the higher price at the stadium can be lower than Uber's because supply arrives before demand does.

The metric that matters then shifts from per mile to per hour. The airline seat analogy helps. An unused seat on a flight is lost revenue for that time window, and so is an unused Cybercab minute. Two 10-mile trips, one 12 minutes and one 45 minutes in traffic, have the same mileage but very different cost in tire wear, energy, and time. If cost per mile can be pushed to the mid-20s of cents, revenue per hour becomes the optimization target. That is why the fleet brain matters as much as FSD; Elon Musk's line about Grok orchestrating the swarm points to the same integration of positioning and pricing.

Scale adds layers. In the early phase a 500-vehicle Tesla fleet in Austin still leaves peak demand to Uber's human drivers. In the second phase private owners opt in with thresholds: 'add my Model Y if it earns $15 per hour,' 'only 9 a.m. to 3 p.m.,' or 'only if it clears $100 per day.' In the third phase large owners with 100 Cybercabs each join as fleet partners. The software must then optimize a mixed fleet of company, private, and partner assets as one network, or both price and availability break.

Tesla has rehearsed this with Supercharging. Prices already move across three daily tiers with demand, not hyper-dynamic but directionally the same. Robotaxi needs the same logic at much higher frequency and granularity, recomputed for every request. The guest's remark that fleet-optimization software may be as important as FSD from a business standpoint captures the point: the best driving is not enough without the best placement.

Subscriptions: Not Count, But Revenue

The subscription chapter starts by separating count from revenue. Ranked by count, Microsoft at 450 million, Google at 350 million, and Netflix at 301 million lead today's tables. Tesla FSD at 1.4 million and Starlink at 12 million sit near the bottom. The video's table, drawn from Tesla Economics' post, is indicative rather than exhaustive, and the debate in the comments is expected.

Rank by annual revenue per subscriber and the order flips. Verizon averages about $732 per year, Netflix about $170, Microsoft about $220. FSD at $99 per month is $1,188 per year, likely the highest on the list, and at $200 per month it would be even higher while also bundling insurance value in some markets. At that level 83 million FSD subscribers would equal Microsoft Office's annual revenue, which is why the guest frames the comparison as count times price. People cancel Netflix, the line goes, nobody cancels transportation.

The forward projection is intentionally round and large to show direction, not a dated forecast. One billion Optimus robots at $1,000 per month is $12 trillion per year, 500 million digital helpers at $420 per month is $2.5 trillion, 300 million Robotaxi subscribers at $500 per month is about $1.8 trillion, Starlink at 600 million subscribers at $67 per month is $0.5 trillion, and FSD at 125 million subscribers at $99 per month is $150 billion. Today's largest subscription revenues at Microsoft and Apple sit near $99 billion each; the projected Tesla and SpaceX pools dwarf them if realized. Timing is uncertain, but the vector is clear: if labor, intelligence, transportation, and connectivity all become subscriptions, Tesla and SpaceX play in four giant markets at once.

Payments close the loop. Per-ride credit-card fees weigh more on small tickets; a once-a-month subscription dilutes the fee over a larger basket and saves Tesla money at scale. Integration with X Money could remove the fee entirely. Bundling follows naturally: Grok, entertainment, connectivity, and rides in one plan, with Robotaxi offering priority or discounted miles the way Amazon Prime offers shipping. An ad-supported free ride is floated as an opt-in, default ad-free but available for price-sensitive riders who trade attention for fare. The goal is not to beat Uber on a single surge, but to make ownership feel optional.

Three open questions remain in the final frame. On rules, the federal direction has turned positive with a performance lens, but until the five FMVSS amendments and new ADS performance tests are finalized, Tesla's certification claim remains legally fragile and Texas permission does not preempt federal vehicle law. On pricing, dynamic network pricing can undercut surge games, but robustness in snow, rain, fleet faults, and cross-state regulatory splits will be tested in the field. On subscriptions, the math is compelling, but the path from $99 to $200 per month depends on unsupervised capability that customers trust enough to keep, and projections stay projections until renewal proves them.