Section 10
Risks & What's Contested
A consolidated, attributed view of what could go wrong — and the genuinely open questions on which informed observers disagree.
9 sourcesAs of 2 June 2026
The risks are not hypothetical — most are already visible in the 2025 numbers: fading credits, a demand cliff after the EV tax credit expired, Cybertruck’s struggles, a regulatory probe covering 3.2M vehicles, and a ~400× valuation that leaves no room for disappointment [14][73][44][57]. The contested part is how much each one ultimately matters.
The concrete risks
Demand and incentives
The US federal $7,500 EV tax credit expired Sept 30, 2025, pulling demand forward and forcing Tesla into its most aggressive incentives yet — 0% APR financing and $0-down leases [73]. Demand support is being withdrawn just as competition intensifies.
Cybertruck
Tesla’s halo product has badly missed expectations: it sold just ~6,406 units in Q1 2025 (a 50% sequential drop), drew a ~46,000-unit recall over a trim panel that can detach in motion, and depreciates ~35% in a single year [75][74][77]. In Q4 2025, SpaceX bought 1,279 Cybertrucks — ~18% of US registrations — without which registrations would have fallen ~51% [76].
Regulatory and safety
NHTSA escalated its FSD low-visibility investigation to an Engineering Analysis covering ~3.2 million vehicles (nine incidents, one fatality) — the final step before a possible recall — alongside a separate robotaxi inquiry [44].
Valuation
At ~400× trailing earnings, the stock prices in autonomy and robotics success that has repeatedly slipped; any delay or safety setback to the robotaxi/FSD narrative could compress the multiple sharply[57][50].
✅The mitigants are real too
Tesla isn’t without defenses. It
returned to delivery growth in Q1 2026, its
energy business is booming at high margin, and in a sign the product itself is solid, Tesla
cracked Consumer Reports’ reliability top 10 for the first time in 2025 — though partly because its lineup is aging
[78]. A genuine robotaxi breakthrough would reframe every risk above.
What reasonable people disagree about
The optimist reads it as
- +A temporary trough between the EV S-curve and the autonomy S-curve, with energy carrying the gap [54].
- +A brand dip that fades once Musk re-focuses and new products (incl. a real affordable model) arrive [79].
- +A data and vertical-integration lead that still converts into the cheapest path to scaled autonomy [35].
The skeptic reads it as
- −A structural decline: two years of falling deliveries, a lost crown, fading credits and no affordable car yet [14][18].
- −A brand and key-man problem that competition will compound, not relieve [69].
- −A valuation detached from fundamentals, exposed to any autonomy delay or regulatory action [44][50].
The weighing
Having laid out both sides, here is where the evidence actually leans on each decisive question — with the concrete tripwires that would change the reading.
On whether Tesla is a car company in decline or an AI company in disguise: the evidence leans toward a real decline in the car business with an unproven AI business on top (medium confidence). The controlling evidence is two consecutive annual delivery declines to 1,636,129 (−9%) [51] and the first-ever revenue decline with net income down 46% [52], which outweighs the Q1 2026 rebound (+6.3%) because that quarter also built ~50,000 more cars than it sold, against a weak comparable[56]. The strongest surviving counter-argument: energy storage is a genuine second act — a record 46.7 GWh (+48%) at roughly double automotive margin, already near a quarter of gross profit[54][13]. What would flip this reading: full-year 2026 deliveries back above the 2023 peak of ~1.81M [51] in the January 2027 delivery report, or energy revenue sustaining>25% growth through the Q4 2026 update. Pre-mortem: if this looks wrong in two years, the most likely reason is treating a lineup-and-politics trough as structural decline — or, on the other side, letting the energy ramp and one rebound quarter obscure a core franchise still ceding share.
On whether the autonomy bet is real or perpetually late: the evidence leans toward real technology that is chronically late against its own promises (high confidence on the lateness; contested on eventual conversion). The controlling evidence is that FSD remains Level 2 with unsupervised driving pushed to “probably” Q4 2026 [41] and a robotaxi pilot running ~42 cars at 19% availability against a promised 500 [40], which outweighs the 10-billion-mile data lead [35] because a decade of mileage accumulation has never yet converted into Tesla removing the driver and accepting liability — while Waymo already runs ~500,000 driverless rides a week [43][36]. The strongest surviving counter-argument: 10B+ miles compounding at ~29M a day, plus a vision-only cost structure that, if it works, deploys across millions of existing cars far cheaper than a sensor-heavy fleet[35][43]. What would flip this reading: Tesla operating without safety monitors and bearing driving liability across the Austin fleet by Musk’s own Q4 2026 date [41]; conversely, a recall ordered out of NHTSA’s 3.2M-vehicle Engineering Analysis would harden the skeptical read [44]. Pre-mortem: if this looks wrong in two years, the most likely reason is that the supervised pilot was the last mile before a fast unlock — or, on the other side, that “late” quietly became “never” while the stock kept paying for it.
On whether ~$1.5T is froth or foresight: the evidence leans toward a price that embeds autonomy outcomes the operating record does not yet support (medium confidence; explicitly not a call on the stock). The controlling evidence is a ~$1.56T market cap on $3.79B of profit [57][52] with robotaxis — under 1% of revenue — driving over $1T of the value [50], which outweighs the ARK and Wedbush cases [48][49] because those are projections conditional on exactly the conversion the autonomy record shows slipping. The strongest surviving counter-argument: the market has repriced real news fast — Musk’s DOGE retreat alone added ~$158B in days — and the high-margin energy business is compounding underneath [59][54]. What would flip this reading: trailing earnings rebuilding toward the ~$15B of 2023 [52] in the FY2026 10-K (early 2027), so the multiple compresses through growth rather than price; regulatory credits actually hitting ~$0 by 2027 with auto margins holding would also retire the subsidized-profit critique [14]. Pre-mortem: if this looks wrong in two years, the most likely reason is unsupervised FSD shipping and recategorizing the earnings base — or, on the other side, the multiple grinding down toward the auto business Morgan Stanley values at ~$55/share [61].
On whether Musk is the irreplaceable asset or the central risk: contested — and it is the evidence itself that deadlocks it, because both sides cite the same dependence: the board calls him critical and shareholders re-endorsed him by >70% even excluding his own shares [65][67], while the same concentration is what large holders flag as key-person risk [66]. On the narrower question of whether his politics damaged the car business, the evidence leans yes (high confidence): the brand fell from #8 to #95 of 100 [69], brand value dropped ~36% in 2025 [81], and Tesla fell ~28% in a European market where overall BEV sales grew ~30% [22][8] — share loss that lineup age alone struggles to explain, though analysts are right that lineup and competition compound it [72]. The strongest surviving counter-argument: the market prices him as a net asset — his refocus on Tesla added ~$158B of value in days [59]. What would flip this reading: European deliveries recovering meaningfully in the full-year 2026 country data (January 2027); Tesla climbing back up the 2026 Axios/Harris reputation poll. Pre-mortem: if this looks wrong in two years, the most likely reason is underrating how fast brand damage fades once political attention moves on — or, on the other side, treating key-person risk as theoretical right up until succession is forced.
That is the weighing; the evidence behind every line of it is laid out in the sections above. See Methodology & Limits for what is disclosed, what is estimated, and where this analysis could be wrong.