Product5 publishers3 min readPublished
The EU's type approval cap puts Bolt's 25,000 driverless Lucids seventeen years out
Bolt aims to put at least 25,000 driverless Lucids on European roads under a non-binding framework with no order and no date, while the only EU-wide approval route for a fully automated car allows 1,500 of a type a year.
The Product Desk · Product desk
What happened
- Bolt aims to deploy at least 25,000 fully autonomous vehicles built on Lucid's upcoming mid-size EV architecture, designed for SAE Level 4 and expected to use Nvidia's Hyperion compute and sensor suite.
- No money has changed hands, Bolt has not placed an order, there is no public deployment timeline, and the companies did not say who will supply the autonomous driving software.
- The only EU type approval route that works across the whole union for a fully automated car is capped at 1,500 units a year per vehicle type, and national approvals stop at each border.
Compiled by The Product DeskSomething wrong?How this is made
Why it matters
- constraint Getting to 25,000 on one approved type takes about seventeen years at the EU-wide rate, so the number depends on the cap changing or on a country-by-country patchwork of national approvals.
- decision Bolt is writing vehicle, software and safety requirements for a driving system that has no named supplier. The choice of autonomy vendor comes after the choice of vehicle.
- precedent Two Lucid deals now have the carmaker supplying hardware while autonomy comes from someone else, and an unordered, unpaid fleet total is becoming the standard unit of robotaxi announcement.
Europe has one driverless taxi service running with nobody in the car: a 22km route in Zagreb, cleared under Croatian rules this month, booked through Uber, operated by Verne on Pony.ai software in a Chinese-built crossover [12]. InsideEVs counts two companies that have deployed driverless taxis in Europe, and one that has approval to drop safety monitors [13].
Type approval binds the number before factory capacity does. The only type approval path that works across the whole union for a fully automated car allows 1,500 units a year per vehicle type [7]. Divide 25,000 by that and a single approved type needs about seventeen years [9]. Bolt's stated ambition is 100,000 autonomous vehicles on its platform by 2035 [11]; nine years at 1,500 a year gets one type to 13,500 [10]. National approvals are the alternative, and they stop at each border [8].
Lucid supplies the car. Nvidia's Hyperion supplies the compute and the sensor suite, and the announcement says the partners will work with autonomous-driving technology partners for the system that actually drives [3][4]. In the Uber deal, that partner is Nuro [19]. The companies did not name one here, and Lucid declined to comment beyond the press release [5][26].
Bolt Autonomous Driving Solutions would define the vehicle, software and safety requirements, build the fleet infrastructure and city partnerships, and own and operate the cars [25]. InsideEVs points out what Level 4 means for that owner: if something goes wrong, responsibility falls on the carmaker or the fleet manager [30]. TechCrunch reported that no money has changed hands and Bolt has not placed an order, and the framework is non-binding [5][6].
On the supply side, Lucid delivered 3,953 vehicles in the second quarter [17]. Its two announced fleet deals total 45,000 cars [20], which is about eleven quarters of that output [21]. The mid-size vehicle those 25,000 depend on slipped by almost a year, to the second half of 2027 [14]. Silvio Napoli became permanent CEO on June 1 [18]. The company cut 18% of its workforce, about 1,500 jobs, later that month, and withdrew its 2026 production guidance [15][16].
Bolt founder and CEO Markus Villig said "autonomous driving in Europe requires data, software, vehicles, and operations to work as one system built for European roads and regulation" [23]. Napoli said Bolt's "reach and operating expertise make it an ideal partner to scale autonomous mobility across Europe" [24]. He has also said the robotaxi business would carry margins that "vastly" exceed those of "the traditional retail model" [22].
A fleet number only tells you about demand if the vehicle has been ordered and paid for, the company writing the driving software is named, and the approval route can carry the volume. Here the vehicle is not ordered or paid for, the software company is not named, and the route allows 1,500 a year per type [5][7]. What Bolt holds is a free option on a vehicle that does not exist yet. For an operator with 850 cities of dispatch data [29] and no capital at risk, that is a sensible position. Lucid carries the schedule, and its first robotaxi, a Gravity running Nuro's driver, is supposed to launch by the end of 2026 [27].
What to watch
- Whether Bolt names an autonomous driving software supplier or converts the framework into an order with money attached.
- Any EU move to raise or replace the 1,500-per-type annual cap on fully automated vehicle approvals.
- Waymo's stated target of Munich in late 2027, which would put a driverless service in a large EU market first.