TSLA · Forward model · FSD / Autonomy · Bear case
What has to happen in FSD / Autonomy
Model as of
This page changes FSD / Autonomy inside the complete TSLA model. Every other vertical, corporate programme and valuation assumption stays in the selected portfolio case.
Shares this vertical and portfolio case. Slider and horizon edits stay in your browser.
FSD / Autonomy
Basis quarter$600M
Final quarter$1.15B
Implied CAGR+14%
Final revenue mix4%
The software layer sold on top of the fleet: one-time purchases and subscriptions. High margin, attach-rate driven, and the enabler without which Robotaxi does not exist.
Last four quarters
2025 Q3
$450M
Estimated
2025 Q4
$500M
Estimated
2026 Q1
$551M
Estimated
2026 Q2
$600M
Estimated
FSD subscriptionsOne-time FSD purchasesLicensing to other OEMs
Subscribers
1.7M
28.3% of a 6.0M addressable base
1.7M paying subscribers against a 6M HW4-capable fleet — 28% attach, and the whole opportunity.
Addressable subscribers
6.0M
the S-curve ceiling
6M HW4-capable vehicles. Older hardware cannot run current FSD and has never been retrofitted at scale.
Net adds
0/qtr
ramping toward 0/qtr, throttled as the base approaches the TAM
Net-add ceiling
0/qtr
what supply can deliver at full rate
ARPU
$99.00/mo
drifting 0.0% per quarter, floor $79.00
$99 a month, the current subscription price. Held flat: Tesla has cut this price before, never raised it.
Non-subscriber revenue
$95M/qtr
growing +8.0% per quarter
$100M a quarter of deferred one-time FSD purchases recognised, plus early OEM licensing.
FSD / Autonomy
Latest: $1.15B (2031Q2E)
| Period | Value |
|---|---|
| 2025Q2 | $380M |
| 2025Q3 | $450M |
| 2025Q4 | $500M |
| 2026Q1 | $551M |
| 2026Q2 | $600M |
| 2026Q3E | $720M |
| 2026Q4E | $821M |
| 2027Q1E | $904M |
| 2027Q2E | $973M |
| 2027Q3E | $1.03B |
| 2027Q4E | $1.07B |
| 2028Q1E | $1.11B |
| 2028Q2E | $1.14B |
| 2028Q3E | $1.16B |
| 2028Q4E | $1.17B |
| 2029Q1E | $1.18B |
| 2029Q2E | $1.19B |
| 2029Q3E | $1.19B |
| 2029Q4E | $1.19B |
| 2030Q1E | $1.19B |
| 2030Q2E | $1.18B |
| 2030Q3E | $1.18B |
| 2030Q4E | $1.17B |
| 2031Q1E | $1.16B |
| 2031Q2E | $1.15B |
Assumptions & reasoning
- Subscribers penetrate a GROWING fleet toward a 90% attach ceiling, so growth slows for two reasons in sequence: first as attach saturates the cars already sold, then because further growth can only come from new deliveries. Once penetration flattens near 90% this line tracks vehicle sales, which is what the automotive vertical above is already modelling.
- TAM is the HW4-capable fleet, 6M at the basis quarter growing 4.9% a quarter, not the whole installed base: older cars lack the compute and retrofitting has never been offered at scale. Raise the growth rate if you think hardware upgrades become a product.
- Margin is 88% because this is software delivered over an existing connection to hardware the customer already bought. Capex is nominal — the training compute is carried at the corporate level, not here.
- The load-bearing assumption is regulatory, and it is not in any slider on this page: the ramp encodes approvals arriving steadily. A single jurisdiction blocking unsupervised indefinitely would break the ceiling, not the ramp.