Few initiatives at any public company carry as much weight—or as much controversy—as Tesla's Full Self-Driving program. For bulls, FSD represents the gateway to a software-defined future: recurring revenue at SaaS-like margins, a burgeoning robotaxi platform, and a fundamental re-rating of Tesla's valuation. For bears, it is a sprawling liability: billions in deferred revenue, serial timeline failures, mounting class-action litigation, and a customer base that increasingly feels betrayed. Both views are supported by evidence in roughly equal measure, making FSD one of the most consequential—and most binary—variables in the Tesla investment thesis.
This report examines the FSD ecosystem across four critical dimensions: the strategic pivot to subscription monetization, the technology's actual capability trajectory, the legal and regulatory exposure accumulating on the balance sheet, and the geographic and competitive headwinds that will shape whether Tesla can reach its audacious 10-million-subscriber target.
The Subscription Pivot: From Perpetual License to Recurring Revenue
A Radical Restructuring of Monetization
Tesla has fundamentally altered how it charges customers for what it calls Full Self-Driving capability. Historically, FSD was offered as a one-time, perpetual license, with pricing ranging from $4,000 to $15,000 depending on the purchase era 3,13,14,19,44. That model has been progressively de-emphasized—and in many markets, replaced entirely—by a $99-per-month subscription in the United States 1,42,48,52, with the European equivalent priced at €100/month 42. The U.S. subscription price point is among the most heavily corroborated claims in the dataset, cited by three independent sources 1,42,52.
Social media reports allege Tesla has gone further, requiring an FSD subscription on new vehicle deliveries and discontinuing the one-time purchase option altogether 15. These claims, however, originate from individual social media posts and lack broader corroboration at this stage.
The Strategic Rationale
The logic behind this pivot is straightforward and well-documented across multiple claims. Subscriptions generate "stickier" revenue streams 25, insulate Tesla from the volatility of quarterly delivery cycles 25, and improve overall revenue quality through higher-margin recurring flows 25. Tesla is actively streamlining the subscription signup process to reduce friction 27 and has introduced gamification features—including streak tracking and detailed usage statistics—to boost engagement and retention 16,22,27. The gamification push is notably well-corroborated, with three independent sources confirming its deployment 16,27.
The logic extends beyond mere retention. Higher FSD usage generates more real-world driving data, which in turn trains Tesla's neural networks 16, creating a virtuous cycle: more subscribers → more data → better performance → more subscribers.
The Risks of Forced Adoption
Yet the forced-subscription approach carries material hazards. Claims suggest it could erode customer satisfaction and degrade the purchase experience 15, attract heightened scrutiny from analysts and press 15, and provoke competitive responses from rival automakers offering different pricing models 15. The annualized subscription cost of roughly $1,200/year 39 compares unfavorably to the one-time purchase for long-term owners, and some owners who previously paid $8,000–$15,000 for FSD outright now face a product whose terms and transferability have been altered after their purchase 2,51.
Executive Compensation and the 10 Million Subscriber Target
Perhaps the most important structural detail is how deeply FSD adoption is now embedded in Tesla's executive compensation framework. Tesla's proxy statement ties CEO compensation to FSD-related metrics covering both up-front sales and subscription revenue 31, with a specific target of reaching 10 million active FSD subscriptions by 2035 encoded in Elon Musk's compensation package 28. This target is corroborated across multiple claims and is remarkable for its ambition.
Current FSD users are estimated at roughly 1.28 million 4,37, meaning Tesla must grow its subscriber base by approximately 8 times over the next nine years. Critically, the compensation metric explicitly excludes free trials 31, ensuring the target reflects genuine paid adoption. FSD subscriber growth also appears as a metric in broader executive compensation targets 46.
This alignment creates powerful incentives for Tesla's leadership to prioritize FSD adoption above nearly all else—but also raises uncomfortable questions about whether aggressive subscription-pushing tactics are being driven by compensation targets rather than genuine customer demand.
Technology: Improving Substantially but Far from Autonomous
The Architecture Evolution
The technical trajectory of FSD is a study in genuine progress constrained by fundamental limitations. Tesla transitioned from a hard-coded C++ approach in FSD v11 52 to a neural-network end-to-end architecture beginning with v12 around June 2025 52. The underlying hardware has also evolved, with HW4 shipping since January 2023 28 and memory per processor doubling from 16 GB to 32 GB 2.
The scale of Tesla's data advantage is significant. The fleet has accumulated approximately 4 billion cumulative FSD miles 46, with Elon Musk claiming close to 10 billion 2—a figure corroborated by two sources—and daily fleet usage reportedly around 20 million miles 43.
User Testimonials and Fleet Metrics
Individual user reports paint an encouraging picture at the anecdotal level. One user drove over 6,000 miles with 97% FSD usage 46; another completed a 700-mile trip with minimal interventions 46; and commenters claim FSD v14 handles "well over 99.5%" of driving scenarios 47. Each neural network in the vision-only stack processes gigabytes of data per second 26, and the system's performance has clearly improved across successive versions.
The Supervised Ceiling
Yet despite these impressive metrics, the system remains firmly in the "supervised" category—and the gap between supervised and unsupervised autonomy remains wide.
Critical disengagement metrics for v14 initially showed improvement toward roughly 2,000 miles between critical disengagements but subsequently regressed to approximately 1,000 miles 3. Regular disengagements in urban driving still occur every few dozen miles 3. Users report a catalogue of specific failure modes including unsafe lane changes into short bypass lanes 8, degraded performance in rain, snow, and low-angle sun 8, phantom braking, multi-minute vision system crashes 34, speed-limit recognition errors 34,35, and even incidents of vehicles driving into lakes 47.
The system is not listed on California's autonomous vehicle deployment list 18, and Musk's most recent promise of unsupervised FSD is targeted for "probably Q4" 2026 12. This timeline must be viewed against a well-documented history of repeated delays: a 2019 claim of feature-completeness by year-end 3, a 2017 promise that drivers could sleep in their cars within two years 49, and numerous other slipped commitments 5,10,12,50.
Musk himself has acknowledged that unsupervised FSD and robotaxi revenue would "not be super material" in the current year 17, and FSD v15 is expected by the end of 2026 7. Explainability features in the neural network stack are often disabled in production due to compute constraints 30, making it difficult to diagnose and remedy failure modes systematically.
Legal and Regulatory Exposure: A Material Overhang
Class Action Litigation
The legal landscape surrounding FSD represents one of Tesla's most significant contingent liabilities—and one that is arguably underappreciated by the market.
A class action lawsuit concerning FSD was recently allowed to proceed 11,38, creating both legal and reputational risk with potential refund obligations 11. Tesla faces up to $14.5 billion in total lawsuit exposure related to Autopilot and FSD 9, and in a tail scenario, could be required to refund all FSD purchases plus punitive damages 19. The company carries billions in deferred revenue on its balance sheet tied to FSD sales 6, representing material exposure if the technology fails to deliver on its marketed promises.
Regulatory Investigations
Regulatory scrutiny is multi-pronged and escalating. NHTSA's engineering analysis covers 3.2 million vehicles 3, an active investigation that could result in a recall 29. The California DMV issued a corrective-action notice regarding FSD marketing claims 54, finding that Tesla's public representations may have overstated the system's capabilities.
Tesla's own Vehicle Safety Report compares FSD miles to national crash averages 6, but critics argue this comparison is misleading because FSD is used predominantly on inherently safer highways rather than the mixed urban environments that generate most accidents 6.
Customer Grievances as Legal Fuel
Customer complaints compound the legal risk and provide a rich evidentiary record for plaintiffs. Owners report paying thousands for FSD without receiving equivalent capabilities across different hardware generations 36,51, being charged the same subscription price for inferior HW3 performance 43,51, and losing FSD transferability upon vehicle resale 2,51.
Broad customer sentiment reflects "skepticism, anger, and feelings of being misled" 52, with active refund demands 38,52. Tesla's 2022 Model 3 sales page promised "full self-driving capabilities in the future—through software updates" 13, language that plaintiffs may leverage to argue that Tesla sold a product it knew it could not deliver within a reasonable timeframe.
Geographic Expansion Challenges
FSD's international rollout faces structural headwinds that could constrain the path to 10 million subscribers. EU certification requirements limit Tesla's ability to release new FSD versions as frequently as it does in the U.S. 45, and the subscription service is not yet available in every European country 33. FSD availability differs between EU markets and Australia 53, and UK availability has lagged the U.S. 52.
There are also technology-readiness concerns about whether FSD Supervised can handle the diverse road conditions, traffic patterns, and regulatory environments across European markets 23. Tesla markets "FSD Supervised" as a distinct product in the EU 18, and the Netherlands—with an estimated 6,000 subscribers 51—provides an early case study. One Dutch owner reported being charged €50 immediately upon subscribing despite not receiving the corresponding software update 24, highlighting operational challenges in the rollout.
The Total Cost of Ownership Context
The FSD subscription sits within a broader ownership cost structure that is worth understanding for context. According to The Zebra data cited by CNBC Select, average annual insurance costs for Tesla vehicles range from $3,153 for a Model 3 20 to $4,156 for a Cybertruck 20 and $4,032 for a Model S 20, with the Model X averaging approximately $377/month, or roughly $4,524/year 20. Owner-reported figures vary from $1,200–$1,500/year 40 to an extreme of $600/month for two vehicles 40.
Adding a $99/month FSD subscription ($1,188/year) to these insurance costs meaningfully increases the recurring expense burden. The total cost of ownership for a Tesla with FSD hardware is estimated at $30,000–$50,000 or more versus car-service alternatives 32. Additional subscription layers compound the picture: Premium Connectivity at $10/month 41, Supercharger Membership at $12.99/month 21, and night charging plans at $15/month 31.
Analysis: The Binary Outcome Embedded in Tesla's Stock
The FSD cluster reveals Tesla at a genuine inflection point. The company is attempting to transform from a cyclical automaker into a software-and-services platform—a transition that, if successful, would justify a dramatically higher valuation multiple. The subscription economics are compelling on paper: 1.28 million users at $99/month implies roughly $1.5 billion in annualized subscription revenue, and the 10 million subscriber target by 2035 would represent $12+ billion in high-margin recurring revenue.
However, the path from here to there is fraught with execution risk. The technology remains supervised-only, with meaningful failure modes that preclude the unsupervised autonomy required for robotaxi deployment. The legal overhang—potentially $14.5 billion in exposure—represents a material contingent liability that could crystallize if class actions succeed. Customer trust has been eroded by years of unfulfilled promises, hardware incompatibility surprises, and changing transfer policies. And international expansion, critical for reaching 10 million subscribers, faces regulatory friction that limits Tesla's ability to iterate at its preferred velocity.
The gamification strategy 16,27 and simplified subscription flows 27 suggest Tesla recognizes it has an engagement problem—many FSD purchasers were using the feature infrequently 27. These tactical moves may improve near-term metrics but do not address the fundamental question of whether the technology can achieve the reliability needed for unsupervised operation.
For investors, the FSD program represents a binary-like outcome embedded within the stock: extraordinary upside if autonomy is achieved and subscriptions scale, but significant downside from litigation, deferred revenue unwinding, and reputational damage if it does not. The alignment of executive compensation with FSD subscriber growth 28,31 ensures management will pursue this path aggressively—for better or worse.
Key Takeaways
-
Subscription economics are compelling, but adoption is nascent. At roughly 1.28 million users and $99/month, FSD subscriptions generate meaningful but not yet transformative revenue. Reaching the 10 million subscriber target tied to CEO compensation requires roughly 8x growth by 2035, demanding both technological breakthroughs and successful international expansion against regulatory headwinds.
-
Legal and regulatory risk is material and escalating. With a class action allowed to proceed 11, up to $14.5 billion in total lawsuit exposure 9, billions in deferred revenue at risk 6, and ongoing NHTSA investigations 3,29, FSD-related contingent liabilities represent a significant and underappreciated balance-sheet risk.
-
The technology is improving but remains far from unsupervised autonomy. Despite impressive fleet-scale data accumulation (~10 billion miles 2) and positive user testimonials, critical disengagement rates, documented failure modes, and Musk's own admission that robotaxi revenue will not be "super material" near-term 17 suggest the gap between current capability and Level 4/5 autonomy remains substantial.
-
Customer trust is a strategic vulnerability. Years of missed timelines 12,50, changing transfer and pricing policies 15,51, hardware compatibility surprises 36,51, and broad sentiment of feeling misled 52 create churn risk that could undermine the subscription growth thesis—particularly as competitors enter the ADAS market with alternative pricing models 15.
Sources
1. Tesla changes FSD transfer rules again, screwing over Cybertruck AWD buyers - 2026-03-04
2. Tesla announces HW4 Plus with doubled memory - 2026-04-23
3. Tesla FSD v14.3 launching this week, Musk claims 'last piece of the puzzle' - 2026-04-01
4. Tesla's Q1 2026 revenue hits $22.38B, driven by EV sales and a 51% surge in FSD subscriptions to 1.2... - 2026-04-23
5. Candid admission by Felon Musk on further $TSLA FSD / robotaxi rollout delays due to safety issues. ... - 2026-04-23
6. Musk falsely claims Tesla FSD is 10X safer than humans, complains about lawsuits - 2026-04-08
7. Musk confirms FSD v15 for AI4 as Tesla pushes for unsupervised driving. #tesla #fsd [Link] Tesla li... - 2026-04-23
8. Musk says Tesla FSD v15 will 'far exceed' human safety - 2026-04-09
9. Tesla is facing up to $14.5 billion in lawsuits - 2026-04-17
10. Elon Musk pushes unsupervised FSD for consumer Teslas - 2026-04-22
11. Na, dann ist zumindest Geld bei #Tesla zu holen, wenn sie Ex-Kunden das Geld für #FSD / "Fake Self-D... - 2026-04-23
12. Tesla confirms Cybercab production has started despite delays in unsupervised driving - 2026-04-23
13. Musk: HW3 can't achieve unsupervised FSD - 2026-04-22
14. Tesla will build factories just to retrofit millions of HW3 cars it said could do FSD ->Electrek | M... - 2026-04-23
15. This does not mean anything. FSD subscribtion is forced on new deliveries for several months. They a... - 2026-04-22
16. Tesla is adding streak tracking and usage stats to Full Self-Driving, using gamification tactics lik... - 2026-04-22
17. Tesla’s Cybercab goes into production — so why is Musk tapping the brakes? - 2026-04-24
18. Tesla gets FSD Supervised approved in the Netherlands - 2026-04-11
19. Elon Musk admits millions of Tesla owners need upgrades for true 'Full Self-Driving' - 2026-04-22
20. The best car insurance for Teslas of April 2026 - 2026-04-22
21. Tesla offers 1 year of free Supercharging, claims ~40% premium for non-Tesla EVs - 2026-04-24
22. Tesla adds ‘streaks,’ other stats to track how often drivers use Full Self-Driving software #Technol... - 2026-04-14
23. #Tesla darf sein teilautonomes Fahrassistenzsystem "Full Self-Driving (Supervised)" erstmals in der ... - 2026-04-11
24. Tesla’s FSD Is Finally Approved In Europe. Only In The Netherlands Though. - 2026-04-12
25. Tesla's revenue is climbing again - and it's not just about selling cars - 2026-04-23
26. Tesla Unsupervised FSD: Why Millions of Vehicles Won't Get Full Autonomy - 2026-04-23
27. Tesla is gamifying Full Self-Driving to get more drivers actually using it - 2026-04-22
28. Tesla adds ‘streaks,’ other stats to track how often drivers use Full Self-Driving software - 2026-04-14
29. The Netherlands is the first European country to approve Tesla’s supervised Full Self-Driving - 2026-04-11
30. Don’t question when your tesla pulls a quirky move - 2026-03-29
31. Free Supercharging for a Year if you buy a Model 3 - 2026-04-25
32. Tesla Announces New AI4+ FSD Computer With More Memory and Compute - 2026-04-23
33. Tesla FSD is approved in the Netherlanfs - 2026-04-10
34. Tesla releases FSD 14.3 - 2026-04-07
35. Tesla charging session - 2026-03-31
36. Elon Musk Shares Specs for Tesla's AI6 Chip, Teases AI6.5 - 2026-04-16
37. Tesla announced start of Cybercab production - 2026-04-23
38. Here are the top 7 voted for questions by investors so far for Q1 earnings call next week: - 2026-04-17
39. Polestar Wants Tesla Owners To Jump Ship With A Massive $21,000 Discount - 2026-04-08
40. Honest thoughts about EV ownership after a month of ownership - 2026-04-02
41. Ford’s CEO Says An Affordable Tesla Model 3, Model Y Rival Is Coming - 2026-04-02
42. What are the flaws of the Tesla Model Y (2026 version)? - 2026-04-14
43. Tesla FSD plows through railroad gate, keeps going - 2026-04-10
44. Car Owners Are Revolting Over Tesla’s Self-Driving Promises - 2026-04-20
45. RDW explanation regarding Tesla's European type approval with provisional validity in the Netherlands - 2026-04-10
46. Anyone here who moved from OpenPilot to Tesla FSD? What’s your experience been like? - 2026-04-11
47. Any elders and people, with disabilities, using self driving cars? - 2026-03-30
48. TSLA Q1 Deliveries: The 50,000 Vehicle Elephant in the Room - 2026-04-07
49. Tesla's first-quarter deliveries miss estimates as tax credit expiry weighs - 2026-04-02
50. SpaceX IPO will create fractioning of Musk shareholder loyalty - 2026-04-05
51. Only hw4 got FSD in Netherlands not HW3 - 2026-04-13
52. HW3 FSD v14 update. - 2026-04-22
53. They fully removed now: „In near future, FSD“ and the car doesn’t react anymore to traffic lights!!! EU M3 2022 - 2026-04-03
54. Fsd name changed on older Model 3. - 2026-04-12
