Robotaxi and FSD Monetization Execution Risk

Tesla's entire bull case now hinges on monetizing autonomy at scale, but the gap between the vision and current financial reality is wide. Despite launching its Robotaxi service in Austin in June 2025 and expanding to seven U.S. markets by FY26Q2, the fleet remains small — 380,000 cumulative unsupervised miles across six cities as of FY26Q2. Management has consistently pushed the timeline for autonomous revenue to become "material" to the second half of FY26, but FY26Q2 operating income collapsed 57% YoY to $398M even as revenue grew 26%.

• The transition to subscription-only FSD pricing creates a near-term headwind to automotive margins, which management acknowledged when announcing the change in Q4 FY25.
• FSD attach rates, while improving (55% of North America deliveries in FY26Q2 had FSD at time of delivery), depend on continued software improvement and regulatory approvals in Europe and China — both of which are moving slowly.
• The Cybercab, which is purpose-built to minimize cost per mile and optimize fleet economics, only commenced production in FY26Q2, meaning Robotaxi economics for the foreseeable future rely on higher-cost Model Y vehicles.
• Expanding personal vehicle owners adding their cars to the fleet (the "Airbnb model") is deferred until fleet kinks are resolved — at the earliest in FY26 — delaying a key mechanism to scale the fleet cheaply.
• Waymo, now with a proven commercial operation across multiple U.S. cities, provides a benchmark that investors will compare against Tesla's ramp.

Unprecedented Capital Cycle With Uncertain Returns

Tesla has entered the most capital-intensive period in its history, with CapEx guidance exceeding $25B in FY26 — a near-tripling from $8.5B in FY25. Free cash flow turned negative in FY26Q2. The company is simultaneously funding six new production lines (LFP cells, Cybercab, Semi, lithium refinery, new Megafactory, Optimus factory), AI compute infrastructure (Cortex 2 at Giga Texas), a semiconductor research fab at Giga Texas, and a planned large-scale TerraFab in partnership with SpaceX.

• Management explicitly acknowledged the CapEx cycle will continue for "two to three more years," with solar manufacturing and semiconductor fab investments not yet reflected in the $25B+ guide.
• Tesla is pursuing debt financing — management disclosed working on facilities to borrow up to $30B — layering financial leverage onto a business with FY26Q2 GAAP operating margin of only 1.4%.
• The $2B investment in xAI, and the deepening entanglement with SpaceX (framework agreements, TerraFab partnership, Starlink integration), raise governance questions about capital allocation between entities that all involve Elon Musk.
• Tesla's new risk factor explicitly warns that CEO performance award targets — directionally consistent with Musk's vision — may not align with products or services that would generate sufficient financial returns, and that failure of demand to materialize could lead to underperformance or write-downs.

Optimus Production Ramp Risk

Management describes Optimus as potentially "the biggest product ever" and targets 1M units per year at Fremont within roughly five years. But Optimus 3 production is only beginning to ramp at Fremont in a factory being converted from the Model S/X line, and management has been emphatic that this ramp will be slower than any prior Tesla product because there is no existing supply chain for any component.

• Rare earth magnet export restrictions from China posed a direct constraint on actuator production in early FY25, illustrating how a single geopolitical event can stall the entire program.
• With 10,000+ unique parts, management stated the production rate will be governed by the "slowest, least lucky" component — making planning nearly impossible and investor expectations difficult to anchor.
• The transition from the Model S/X line to an Optimus production line at Fremont takes months just for dismantling and reinstallation, and production start in FY26 is followed by a "very slow initial S-curve," per Musk in Q1 FY26.
• Optimus 4 — targeting 10M units per year — will require an even more vertically integrated supply chain and a still-to-be-announced TerraFab for AI chips. Both are years away.
• Chinese humanoid robot competitors are advancing rapidly. Management acknowledges China will be "the toughest competition by far" given its manufacturing scale and AI capabilities.

Competition and Share Loss in Core EV Business

While Tesla remains the global EV leader in the U.S. and Western markets, the competitive landscape is intensifying. China is the clearest stress test: BYD and domestic Chinese brands have captured a growing share, and Tesla's FY25 vehicle deliveries declined 9% YoY to 1.64M while BYD continued to grow. In Europe, FSD is still pending regulatory approval as of FY26Q2 (Netherlands approval received; EU-wide review still underway), which Tesla management says suppresses both sales and FSD attach rates in one of its largest markets.

• Automotive gross margin, excluding regulatory credits, declined from a peak of ~29% in FY21-22 to 17.8% in FY25 and 16.3% in FY26Q2, as price cuts have consistently outpaced cost reductions.
• Regulatory credit revenue, which was $2.8B in FY24 at near-100% gross margin, dropped 28% to $2.0B in FY25 after U.S. legislative changes eliminated emissions penalties; this near-zero-cost revenue stream is unlikely to recover.
• Battery pack supply emerged as the primary production constraint in FY25 and FY26Q1, capping the ability to fulfill demand even when it rebounds.
• The IRA EV consumer tax credit ($7,500) was eliminated in the U.S. by the One Big Beautiful Bill Act effective Q3 FY25, increasing the effective price of Tesla vehicles for U.S. consumers.
• Tesla's brand faced headwinds in certain markets in FY25 connected to Elon Musk's political activities, and management explicitly acknowledged this impacted demand.

Energy Storage Margin Compression

Tesla's Energy segment was a bright spot from FY23-FY25, with gross margins expanding from near-zero to 29.8% in FY25. However, management has guided for meaningful compression from here. Energy gross margin already fell sharply from a one-time peak of 39.5% in FY26Q1 (inflated by tariff reversals) to 20.4% in FY26Q2, which management described as including a ~$240M warranty true-up but also reflecting structural ASP pressure from competition.

• Management guided Energy segment to normalize "in the mid- to low 20% range" in FY26, down from 29.8% in FY25.
• The primary cost input — LFP battery cells — is still sourced predominantly from China, with tariffs adding over $200M per quarter in FY25. Local U.S. LFP manufacturing (Nevada) only began commissioning in early FY26 and covers a fraction of demand.
• Competition is intensifying from CATL, LG Energy Solution, and other integrators, particularly on price. Megapack 3 will help on cost, but management acknowledged ASPs are already declining.
• Energy deployments are inherently lumpy — FY26Q1 saw an 38% sequential decline to 8.8 GWh before rebounding to 13.5 GWh in FY26Q2 — making margin forecasting difficult.

Chip Supply Chain and AI Infrastructure Bottleneck

Tesla's ability to scale both Robotaxi and Optimus depends critically on AI chip supply. Management has repeatedly stated that chip production — specifically AI logic and memory — is likely to become the primary constraint on Tesla's growth in three to four years.

• There are currently no advanced memory fabs at scale in the United States, and even maximum output from TSMC (Arizona), Samsung (Texas), and Micron may be insufficient for Tesla's FY28-FY29 plans, per Musk.
• The TerraFab initiative — an integrated logic, memory, and packaging fab — is in very early stages. The Giga Texas research fab is a "few thousand wafer starts per month" R&D facility, not a production-scale fab. A large-scale TerraFab involves SpaceX as lead investor and has not been fully announced.
• AI5 chip tape-out was completed in April FY26 and is in volume production at TSMC (Arizona) and Samsung (Texas). But AI5 is being prioritized for Optimus and data centers, not vehicles, because AI4 is sufficient for unsupervised FSD — leaving vehicle hardware upgrades (for the ~30% of the fleet on Hardware 3 that cannot run unsupervised FSD) as an unresolved and potentially costly challenge.
• The Hardware 3 upgrade problem is large: Tesla must set up "micro factories in major metropolitan areas" to replace computers and cameras in these vehicles to enable them to join the Robotaxi fleet, representing a significant capital and operational burden whose scale has not been quantified.