Tesla
EV maker repositioning as an autonomy and robotics company, with the AI thesis still running ahead of AI profits. Q2 2026 (reported July 22, 2026) — revenue $28.2B (+26% YoY) but operating margin only 1.4%, free cash flow negative $1.1B and capex $5.8B (+142%). Active FSD (Supervised) subscriptions reached 1.48M (+56% YoY) after the February 2026 switch to subscription-only, with supervised FSD approved in the Netherlands, Lithuania, Estonia, Denmark and Belgium. Robotaxi runs in seven metros (Austin, Dallas, Houston, Miami, Orlando and Tampa ramping unsupervised; Bay Area with a safety driver). Paid Cybercab rides began in Austin on September 4, 2026, and NHTSA opened an audit of the vehicle's federal safety self-certification that same week. Q3 2026 deliveries (reported October 2) were 486,532 — down 2.1% YoY but well above the ~462K consensus — while 13.7 GWh of storage missed. Q3 earnings are due October 21.
Scenarios
Robotaxi is live in seven metros with Phoenix and Las Vegas in preparation, and Giga Texas lists Cybercab capacity above 125,000 units a year — if unsupervised operation scales, removing the driver rewrites per-mile economics. FSD subscriptions growing 56% YoY, with over 55% of new North American deliveries attaching a subscription, build recurring software revenue, and five European approvals ahead of a reported EU-wide decision in December could open Europe. Fremont's Model S/X lines are being converted to first-generation Optimus production, giving the humanoid program real factory capacity.
The AI narrative is not yet in the numbers — 1.4% operating margin and negative free cash flow in Q2 2026 while capex more than doubled, and Q3 deliveries fell 2.1% YoY with Model S/X discontinued. Robotaxi reached seven metros against a 10+ city ambition, while Waymo already operates in about 15 cities. NHTSA is auditing the Cybercab's self-certification (no steering wheel or pedals) on top of existing FSD probes, and an adverse finding could stall the purpose-built fleet. Optimus has no disclosed production volume, and hand-manufacturing problems have been reported.
Key Factors to Watch
- ●Q2 2026 (Jul 22) — revenue $28.2B (+26% YoY), operating margin 1.4%, free cash flow negative $1.1B, capex $5.8B (+142%)
- ●Q3 2026 deliveries 486,532 (Oct 2) — down 2.1% YoY, ~5% above consensus; storage 13.7 GWh vs 15.9 GWh expected
- ●1.48M active FSD subscriptions (+56% YoY), subscription-only since February 2026; supervised FSD approved in five EU countries
- ●Robotaxi in 7 metros (6 ramping unsupervised in Texas and Florida); paid Cybercab rides in Austin since Sep 4, 2026
- ●NHTSA audit of Cybercab self-certification (opened Sep 2026) is the key regulatory overhang
- ●Training compute on Cortex 1 (>90 MW) and Cortex 2 (>115 MW) in Texas; onsite compute more than doubled in H1 2026
Score History
| Date | Score | Direction | Note |
|---|---|---|---|
| 2026-10-10 | 6.7 | Positive | Text refresh to Q2 2026 (Jul 22) plus Q3 deliveries (Oct 2), dims unchanged. Removed claims not supported by the Q2 2026 update — London/Berlin unsupervised FSD pilots (deck lists supervised approvals in NL/LT/EE/DK/BE only), Dojo as part of the AI stack (team disbanded Aug 2025; compute is Cortex 1/2), 8.2B FSD miles, FSD v14.3 compiler detail, 10+ city robotaxi. Post-Aug-22 signals offset — paid Cybercab rides in Austin (Sep 4) vs NHTSA self-certification audit. Q3 earnings Oct 21 not scored. |
| 2026-03-08 | 6.7 | Positive | Score 6.5→6.7 (are 6→7). are recalibrated: FSD subscriptions growing, Optimus program, Dojo supercomputer; AI is core investment thesis |
| 2026-03-08 | 6.5 | Positive | Score 6.7→6.5 (formula reweight: sb 0.25→0.15, are 0.20→0.25, md 0.20→0.25, dr 0.20→0.25, aam 0.15→0.10) |
| 2026-03-08 | 6.7 | Positive | Score 7.9→6.7 (are 9→6, dr 2→5). External research cross-ref: FSD <2% of $94.8B revenue (are↓); Waymo 250K rides/wk, BYD overtook EV sales, NHTSA investigations (dr↑) |
| 2026-03-08 | 7.9 | Positive | Initial assessment from batch 7 research |
Transportation Peers
Last researched: 2026-10-10
This is research and analysis, not financial advice. Scores reflect AI impact potential, not investment recommendations.