May 15, 2026
On Saturday May 10, Tesla built the final Model S at Fremont, a black Plaid, and the final Model X, a signed Plaid. The Model S production run ran 14 years from the original 2012 delivery; the Model X production run ran 11 years from the 2015 delivery. Combined lifetime production of the two nameplates lands at roughly 750K units. That is the operative number in the used-car market through 2027. Every Model S and Model X built will now appreciate or depreciate against a fixed supply pool with no new units entering the channel.
The 350-unit Signature Edition is the curated exit. The pricing structure was $159K to $160K per car, 250 Model S Plaid and 100 Model X Plaid, and the run sold out in under a week. The Signature Edition delivery event was scheduled for May 12, the same day Musk left for Beijing aboard Air Force One with Cook, Fink, Powell McCormick, Ortberg, Huang, and roughly fifteen other CEOs. Tesla postponed the delivery event to May 20. The trade-off was structurally revealing. Musk’s presence in Beijing negotiating BYD US market access and FSD-China approval mattered more to Tesla than the ceremonial close of its longest-running production line.
Fremont’s pivot is the larger story. The Model S and Model X lines retool for Optimus humanoid manufacturing, which Piper Sandler’s Alex Potter sized in a Monday May 11 note as the upside Tesla bull case currently being valued at zero. Potter projects Tesla FSD subscribers expanding from 470K today to 10.6M by 2030 and 95M by 2038, with Tesla revenue reaching roughly $1T by 2038. At Tesla’s recent $400 share price, Potter argues investors are buying Optimus for free. The Fremont retooling is the production-side test of that thesis. Optimus has to be built somewhere; the company chose its highest-prestige legacy line.
The Used-Market Implications Land in Q3
The discontinuation creates the largest discontinued-premium-EV inventory pool in US used-market history. Roughly 750K combined Model S and Model X units are now in the wild against a fixed-supply pool. The closest historical comparison is the Chevrolet Volt at end-of-production in 2019, but Volt volume was an order of magnitude smaller and the Volt was never a premium product. The Model S and Model X were both premium products that defined a segment. Lucid Air, Mercedes EQS, BMW i7, Porsche Taycan, and the next-cycle Audi e-tron GT will all price against a 750K-unit fixed pool of discontinued Plaid and Long Range inventory through at least 2027.
The collision-repair side absorbs the pivot differently. Tesla parts inflation on the Model S and Model X will accelerate inside 24 months because Tesla now has no business reason to maintain the parts catalog at OE-cost economics on a discontinued nameplate. The ADAS Paradox covered by GCBC in late February for collision-repair claims severity already runs higher on Model S and Model X than on Model 3 and Model Y because of the aluminum body structure and the panoramic glass. Once the parts supply chain transitions from production-line draw to aftermarket-only fulfillment, total cost of repair on a Model S or Model X collision claim rises by a percentage that depends entirely on how Tesla manages its parts division through the wind-down.
The Signature Edition pricing tells its own story about the appetite for the discontinued tier. $159K to $160K on a Plaid that retailed for $109K to $129K in standard configuration is a 23% to 47% premium that the market absorbed in under a week for 350 units. Tesla left margin on the table on the bulk of the Model S and Model X production run; it priced the exit run to capture the willingness-to-pay it knew was there. The same pattern is visible in the Cybertruck RWD discontinuation at 173 confirmed units, where the trim came off the configurator before it could establish run-rate volume.
What Cybercab Has to Deliver on August 8
The Fremont pivot only works if August 8 lands. Tesla’s August 8 Cybercab and Robotaxi event has to deliver production-spec FSD performance against the unsupervised-robotaxi fleet that grew from 9 vehicles in early April to 26 in early May to 39 by mid-May. The progression is real. Wait-time and cancellation reports surfacing in Austin compromise the unit-economics framework Piper Sandler used to value Optimus at zero, because the framework assumes per-ride costs at scale that the current 39-vehicle fleet has not validated.
The Burkland team’s coast-to-coast Cannonball run on FSD V14.3.2 that closed Manhattan to Redondo Beach in 49 hours 55 minutes with zero driver disengagements is the architectural counterpoint. The claim is pending third-party verification with a 7 to 10 day window. If verified, it beats Alex Roy’s January 22 zero-intervention benchmark of 58 hours 22 minutes by 8 hours 27 minutes, a 14.5% time reduction in less than four months of software-improvement velocity. That is the most concrete 2026 evidence Tesla’s vision-only FSD architecture is approaching the conditional-autonomy threshold the August 8 event needs to make Cybercab a production proposition rather than a forward-revenue placeholder.
The competing architecture is the Avatr 07L’s Huawei 896-channel LiDAR and the BYD Seagull’s launch as the first sub-$10K production EV with LiDAR globally. The Chinese cost curve crossed the sub-$200 production-spec LiDAR threshold in 2026, which means premium-segment LiDAR is no longer a price discriminator on a 70,000-yuan A00-class vehicle. Tesla’s vision-only Cannonball claim and Avatr’s 896-channel LiDAR are the two poles of the 2026 ADAS-architecture debate, and they are converging on the same question: which approach gets to L4 first at unit economics that survive a Q3 dealer audit.
The 39-vehicle unsupervised fleet, the 1.3M FSD subscriber base at end-Q1, and the planned H1 2026 expansion to Phoenix, Miami, Orlando, Tampa, and Las Vegas are the operational ramp. The Fremont retooling, the discontinued S and X nameplates, the Optimus production commitment, and the Cybercab event are the financial-narrative ramp. The two have to converge on August 8 for the Piper Sandler “Optimus for free” thesis to hold and for the TSLA -2.6% to $433.45 Tuesday-to-Thursday give-back to look like noise rather than the early read on the FSD-China-approval narrative that traveled to Beijing with Musk.
The Fremont line that just built its final Model S has produced premium EVs continuously since 2012. The bench it just vacated is now the production floor for the bet Tesla is making instead. Whether the bet works will be settled by what comes off that floor in Q3, not by the headlines that closed the line.
