Tesla Q3 2026: I Read the 10-Q Footnotes First
Summary
Tesla's Q3 2026 results came in with total revenue of $28,046M and GAAP diluted EPS of $0.22, down roughly 44% year-over-year, while non-GAAP EPS held at $0.48 — two numbers that seem to contradict each other before you open the footnotes. The central context is an IEEPA tariff refund disclosed in the Form 10-Q: by my math, $526M in energy cost reductions plus $260M in automotive cost reductions minus $339M passed back to energy-storage customers produces a net effect of approximately $447M, equal to roughly 62% of the reported $718M operating income. Remove the refund and operating income falls to approximately $271M; the refund-adjusted gross profit of roughly $5,047M is virtually unchanged from last year's $5,054M. Operating expenses jumped 39.3% to $4,776M, with roughly 35% of that increase — approximately $470M — driven by stock-based compensation tied to the CEO Performance Award, a charge carrying $9.74B in remaining unrecognized balance expected to flow through earnings over about 8.9 years. Strong operating metrics — a 17.0%-to-18.5% automotive gross margin improvement, 32.5% services revenue growth, and FSD active subscriptions rising from 1.04 million to 1.74 million — are real, but reading this quarter accurately requires separating those genuine improvements from the tariff refund recognized this quarter, which dominated the headline income figure.
Key Points
The $447M net tariff refund accounts for roughly 62% of reported operating income
Tesla's Q3 Form 10-Q discloses IEEPA tariff refunds reducing energy cost of revenues by $526M and automotive cost of revenues by $260M, while $339M was deducted from energy revenue as amounts owed back to customers under contract — producing a net effect of approximately $447M by my math. Set that against the reported $718M operating income and you get roughly 62%. Strip the refund out and operating income falls to approximately $271M. The refund-adjusted gross profit, by my calculation, is approximately $5,047M — nearly unchanged from last year's $5,054M. The refund wasn't material in the first half of the year, meaning this Q3 recognition represents a concentrated single-quarter event. Neither GAAP nor non-GAAP adjustments strip the refund out as a separate line item, so the full benefit flows through both EPS figures. Ford disclosed receiving approximately $500M in IEEPA-related refunds in Q3 2026 and expects approximately $2.2B in total, though timing is uncertain and subject to trade-policy changes — confirming this is an industry-wide variable, not a Tesla-specific quirk.
Operating expense increase: roughly 35% came from stock-based compensation
Q3 operating expenses rose from $3,430M to $4,776M, a $1,346M or 39.3% increase, with R&D jumping from $1,630M to $2,761M and SG&A rising from $1,562M to $2,015M, partially offset by the absence of last year's $238M restructuring charge. By my math, R&D SBC went from $335M to $513M and SG&A SBC went from $121M to $413M, bringing total SBC from $456M to $926M — an increase of approximately $470M, equal to roughly 35% of the full OpEx increase. The Form 10-Q attributes the SG&A SBC surge primarily to the 2025 CEO Performance Award, with $275M recognized in Q3 alone; the remaining unrecognized balance is $9.74B, to be recognized over about 8.9 years. R&D spending excluding SBC rose roughly 74% year-over-year by my math, reflecting the scale of non-compensation cash investment in Supervised Intelligence, Optimus, silicon, battery technology, and new vehicle programs. Reading the OpEx increase as entirely reflecting AI investment or entirely reflecting inefficiency misses the real composition: more than a third is equity-based compensation, and the rest is mostly real-dollar project spending that the Form 10-Q breaks down item by item.
CapEx guidance cut to approximately $20B and a $30B credit facility signed
The 2026 CapEx guidance that stood above $25B in the Q2 Form 10-Q was revised to approximately $20B in the Q3 filing; CFO Taneja said on the call, "We're revising our CapEx guidance to be approximately $20 billion for the current year," and Musk said "it's actually ending up being a little lower than we anticipated," but the reason isn't disclosed anywhere in the filing (per the Motley Fool transcript). Nine-month cumulative CapEx stands at $13.01B, which means Tesla needs to execute roughly $7B or more in Q4 alone to hit $20B — a single-quarter figure larger than Q3's $4,732M. On September 29, Tesla entered three credit facilities totaling $30B: a $20B 3-year delayed-draw term loan, an $8B 5-year revolving facility, and a $2B 364-day revolving facility, all undrawn at signing. Free cash flow was $1,531M versus $3,990M a year ago — and was actually negative at -$1,092M in Q2 — while the CFO signaled CapEx keeps rising in future years for Robotaxi fleet expansion. Cash and short-term investments stand at $45.4B with a $5B minimum liquidity covenant, meaning no immediate funding concern, but the $20B delayed-draw term loan's draw timing becomes a meaningful medium-term variable as unused commitments automatically step down starting at the first anniversary of signing.
Services revenue grew 32.5% — but FSD subscriptions belong to a different revenue line
Services and other revenue grew from $3,475M to $4,605M, a 32.5% increase, with primary drivers being higher used vehicle sales volume and average selling price, non-warranty service and collision repair revenue, and increased paid Supercharger usage, per the Form 10-Q. FSD active subscriptions rose from 1.04 million to 1.74 million, a 67.3% increase that combines prepaid and monthly subscriptions and excludes free trials — but the revenue from those subscriptions sits in automotive sales, not services. The Form 10-Q explicitly lists FSD subscription growth as a driver of automotive sales revenue, which means the 32.5% services growth and the 67.3% FSD subscription growth cannot be combined as a single software-transition indicator. More than 100,000 customers in nine EU countries used FSD and drove more than 160 million kilometers, representing early international traction for a feature the company labels as requiring active driver supervision that does not make the vehicle autonomous. Without churn data being disclosed, the long-term durability of the 1.74 million subscription figure as a recurring revenue base cannot be verified from outside. The 1.74 million figure also mixes prepaid and monthly subscriptions, meaning the unit count alone cannot be translated into a software revenue number by outside observers.
Robotaxi and Optimus: business scale is still undisclosed at this stage
Tesla launched commercial Cybercab rides in Austin on September 3, and Exhibit 99.1 shows the entire Robotaxi service accumulated nearly 2 million cumulative autonomous miles — 8x quarter-over-quarter growth — though that covers the full service, not Cybercab alone. As of the earnings call, Cybercab operates only in Austin while Ashok Elluswamy previewed a Dallas Robotaxi launch with Model Y Robotaxis the following month (per the Motley Fool transcript); Musk said Cybercab's fleet grew about 10x in roughly a month but the starting number wasn't disclosed, making the 10x figure unanchored. For comparison, Waymo logged more than 270 million cumulative autonomous miles through end of June 2026 (Waymo blog) — different time reference, much longer operating history, but the scale gap is unmistakable. Optimus is listed as "Early Ramp" in Exhibit 99.1, with no unit volumes or revenue disclosed, and Musk placed external deliveries at least a year or more out, by Musk's own estimate. Neither business discloses weekly trip counts, fleet size, or unit economics, yet both are central to Tesla's long-term valuation narrative, making it difficult to assess the business scale of either from public filings alone.
Positive & Negative Analysis
Positive Aspects
- Automotive gross margin improved from 17.0% to 18.5%
The 150-basis-point improvement in automotive GAAP gross margin is a genuinely positive signal and holds up even when adjusted. The reported gross margin includes the $260M IEEPA refund recognized in automotive cost of revenues, so my math puts the refund-adjusted automotive gross margin at roughly 17.2% — still an improvement from last year's 17.0%, in a quarter where regulatory credits dropped from $417M to $113M. Inventory at roughly 11 days signals tight supply-demand management, and the company attributed the improvement to regional and product mix. Production grew 3.8% from 447,450 to 464,391 units, running ahead of deliveries and preserving room for future delivery recovery. Non-GAAP automotive gross margin excluding regulatory credits also rose, from 15.4% to 18.1%, and achieving that while regulatory credits shrank from $417M to $113M reflects underlying cost-side progress that isn't purely explained by the tariff accounting event.
- Services revenue growing 32.5% builds an expanding post-vehicle revenue stream
The $4,605M services revenue figure represents the only double-digit growth segment in a quarter where automotive and energy revenue both declined. According to the Form 10-Q, the growth was driven by used vehicle sales volume and pricing, non-warranty service and collision repair, and paid Supercharger usage. Services now represent 16.4% of total revenue, and services gross profit was $734M, confirming this segment contributes meaningfully to profitability, not just top-line scale. The Supercharger network expanded 14.8% to 8,904 stations, supporting the trend. While used vehicle sales aren't pure recurring revenue, service and repair revenue and Supercharger usage naturally scale with the growing installed base of about 10.2 million vehicles. Deferred revenue from customer advances also grew from $2.04B at end-2025 to $3.30B, thickening the forward revenue pipeline. The segment is turning into a real and growing profitability contributor.
- Operating cash flow held at $6,263M despite the investment surge
Despite operating income falling sharply and CapEx more than doubling, operating cash flow came in at $6,263M versus $6,238M a year ago — essentially flat. OCF sits above the CapEx line so the FCF story is more complicated, but maintaining cash generation from operations in a quarter of heavy investment and compressed margins is notable. Coming off Q2 with negative FCF of -$1,092M, Q3's positive FCF of $1,531M is a meaningful directional improvement. Cash and short-term investments grew from $41.6B a year ago to $45.4B, showing that even as Tesla invests heavily, its liquidity position strengthened year-over-year. Working capital changes — primarily accounts payable and accrued liabilities — contributed $3,991M to OCF versus $1,646M last year, so that factor warrants acknowledgment in any read of the OCF figure. Deferred revenue growth and the deepening FSD subscription pipeline both provide some forward visibility into future cash inflows.
- Commercial Robotaxi launch and FSD global expansion are real milestones
Launching commercial Cybercab rides in Austin on September 3 marks a transition from testing to live commercial service — a threshold the long-term valuation narrative needed to cross. Cumulative Robotaxi service autonomous miles growing 8x quarter-over-quarter and a Dallas expansion in the pipeline show the service is actively broadening. FSD reaching more than 100,000 customers in nine EU countries and logging more than 160 million kilometers demonstrates the feature is gaining traction in a major international market. The installed base of about 10.2 million vehicles represents a meaningful addressable pool for both FSD subscriptions and future Robotaxi fleet enrollment. The Robotaxi service itself began in June 2025, with Cybercab added in September 2026 — meaning the phased rollout is actually proceeding on schedule rather than slipping. These are directional confirmations that the autonomous strategy is moving from announcement to execution, even if commercial scale remains early and key metrics are still undisclosed.
Concerns
- Strip out the refund and operating income falls to roughly $271M
By my math, removing the approximately $447M net tariff refund leaves operating income at roughly $271M — and the refund-adjusted operating margin at roughly 1.0% on adjusted revenue of approximately $28,385M. The reported 2.6% operating margin already looks weak against last year's 5.8%; the adjusted figure makes the deterioration look sharper. There's no disclosed evidence the refund recurs in Q4; Ford's filing noted timing uncertainty and trade-policy dependence. Gross profit adjusted for the refund is roughly $5,047M — nearly identical to last year — which means the core business hasn't improved its earnings power enough to absorb the OpEx surge on its own. The refund wasn't recognized in the first half and concentrated entirely in Q3, and there's no visible mechanism in the filing to expect repetition. The structural OpEx pressure — including SBC that will flow through GAAP earnings for years — makes the underlying profitability picture more challenging than either the reported GAAP or non-GAAP headline implies.
- $9.74B in unrecognized SBC will press earnings for about 8.9 more years
The unrecognized balance on the 2025 CEO Performance Award is $9.74B, expected to be recognized over about 8.9 years — with $275M recognized in Q3 alone and $802M in the nine-month period. On a simple average, that works out to roughly $1.1B per year going forward, though milestone achievement timing can shift the pace. SBC doesn't consume cash and exits non-GAAP, but on a GAAP basis it erodes EPS every quarter and dilutes existing shareholders through equity issuance. The SG&A SBC line went from $121M to $413M year-over-year — more than triple — and the Form 10-Q attributes it primarily to CEO compensation recognition. The December shareholder meeting's advisory vote on 2025 executive compensation is non-binding and cannot modify the $9.74B balance; it's an opinion signal, not a governance lever that changes the expense schedule. Any forward earnings model for Tesla needs this ongoing expense built in as a structural assumption, and the GAAP-to-non-GAAP gap — which more than doubled from $0.11 to $0.26 per share year-over-year — will persist as long as this recognition schedule continues.
- Free cash flow fell 61.6% year-over-year
FCF dropped from $3,990M to $1,531M — a 61.6% decline driven primarily by CapEx more than doubling to $4,732M. Q2 FCF was negative $1,092M, so Q3's return to positive territory is an improvement in direction, but at roughly 38% of last year's Q3 level, the absolute number remains compressed. With the CFO signaling CapEx keeps rising in future years for Robotaxi fleet and infrastructure expansion, the conditions for FCF recovery aren't immediately visible. The $30B credit facility signed on September 29 — in the same quarter FCF bounced back from negative — suggests the company recognizes its funding needs may extend beyond current cash generation at some point. Cash stands at $45.4B with a $5B minimum liquidity covenant, so there's no near-term crisis, but the Form 10-Q explicitly states the company may need additional funding beyond operating cash flows during periods of high CapEx. Maintaining annual CapEx at $20B or above for several years while FCF stays at or below current levels would accelerate the drawdown pace of the liquidity cushion, making the $30B facility's draw timing a live planning question.
- Robotaxi and Optimus business metrics remain undisclosed
The absence of any disclosed metrics on Cybercab's commercial operations — no weekly trip counts, no fleet size, no unit economics — is the single most uncomfortable uncertainty for investors right now. Musk said the fleet grew 10x in roughly a month, but without the starting count, that figure can't be anchored to any scale. Waymo's more than 270 million cumulative autonomous miles as of end of June 2026 (Waymo blog) versus Tesla's nearly 2 million for the full Robotaxi service — acknowledging different time references and operating histories — puts the scale difference in unmistakable terms. Optimus sits at "Early Ramp" with zero units disclosed and external deliveries at minimum a year away per Musk's estimate. These two businesses are the primary justification for Tesla's long-term valuation premium over a traditional auto multiple, but the public record as of this filing doesn't contain enough data to quantify either one as a business. Until trip counts, fleet sizes, and revenue figures begin appearing in disclosures, the autonomous and robotics thesis remains a narrative that's still waiting for confirming numbers.
Outlook
The most immediately visible near-term milestone is Q4 capital expenditure execution and the December 3 shareholder meeting. With nine-month cumulative CapEx at $13.01B against a revised annual guidance of approximately $20B, Tesla needs to deploy roughly $7B or more in Q4 alone — larger than either Q2's $5,789M or Q3's $4,732M. Whether that actually gets executed matters as a gauge of the company's investment rhythm and the practical credibility of the revised $20B target. Stock price movement after the earnings release is worth noting as context: shares dropped roughly 3.1% to $374.58 on October 22, edged down to $372.22 on October 23, then climbed to $407.46 in the October 26-27 window, and closed November 2 at $403.82. I couldn't identify a single confirmed cause for any specific leg of that two-week trajectory.
At the December 3 meeting at Gigafactory Texas in Austin, three items are on the agenda: election of three directors, a non-binding advisory vote on 2025 executive compensation, and ratification of PwC as auditor. The second item is the one I'm focused on. CEO Performance Award charges of $9.74B in unrecognized balance — $275M recognized in Q3, $802M in the first nine months — will keep flowing through GAAP earnings for years. But the vote is advisory and non-binding; it won't change the outstanding balance or accelerate any expense recognition. I'd be cautious about reading the outcome as a definitive verdict on Tesla's compensation structure or the cost of its transformation. The previewed Dallas Robotaxi launch and Cybercab's expansion to additional cities are also near-term events to watch; how Austin operating hours and geographic scope evolve as data accumulates will be an early signal of service maturity.
The medium-term variable I'm tracking most carefully is the quality — not just the count — of FSD active subscription growth. The jump from 1.04 million to 1.74 million, up 67.3%, is substantial on paper. But the key question is whether this reflects organic adoption expansion in existing markets, or an initial surge from entering new markets like the nine EU countries where more than 100,000 customers drove more than 160 million kilometers. The company's own label matters here: FSD (Supervised) means "Active driver supervision required; does not make the vehicle autonomous." This is a driver-assistance feature, not autonomous driving, and whether people sustain paid subscriptions for an assisted-driving feature over time isn't yet proven by public data. Without churn disclosure, it's genuinely difficult to judge how durable the 1.74 million figure is as a revenue base. The 1.74 million figure also mixes prepaid and monthly subscriptions, so the unit count alone doesn't translate into a revenue number from outside.
Services revenue composition is the other medium-term question that keeps coming up. Q3's 32.5% growth to $4,605M is impressive, but per the Form 10-Q the primary drivers were used vehicle sales, non-warranty service and collision repair, and paid Supercharger usage. Since FSD subscription revenue is classified under automotive sales, you can't use services revenue growth as a proxy for the software transition's pace. If the main drivers remain used cars and physical infrastructure fees, services growth is essentially an extension of the hardware-dependent model. Regulatory credits falling from $417M to $113M year-over-year adds pressure: if that revenue source keeps declining, the need for software revenue to compensate becomes more acute. The lack of granular breakdown in public filings makes external verification hard — and that opacity is exactly what creates investor discomfort. If Tesla begins disclosing services revenue components at a finer level, it would be a significant moment for market assessments of the transition's speed.
The CFO's signal that CapEx will keep expanding over coming years, combined with the $30B credit facility signed on September 29, deserves careful joint reading. Cash and short-term investments at $45.4B look healthy, and the $5B minimum liquidity covenant is comfortably met. But sustaining annual CapEx at or above $20B for multiple years, while FCF stays compressed or dips negative as it did in Q2, means the $30B credit line shifts from a backstop to a potential active funding source. Cortex 1 at 105MW+, Cortex 2 at 120MW+, and a 400MW year-end training compute target — with Austin Fab equipment installation starting in 2027 — represent committed capital before even accounting for Robotaxi fleet expansion, Optimus production ramp, Tesla Semi Nevada production, and battery technology investment. The $20B delayed-draw term loan is the key item to track: it allows up to 10 draws within 18 months of signing, but the undrawn commitment automatically steps down to $10B at the first anniversary and to $5B at 15 months, before expiring at 18 months. Based on the 8-K, the draw window starts shrinking from September 2027 — a closer timeline than it might feel today.
For the long-term valuation thesis to hold, Exhibit 99.1's statement needs to become reality: "over time, we expect our hardware-related profits to be accompanied by an acceleration of SI, software and fleet-based profits." Right now, hardware profits received significant support from a tariff accounting event, while software and fleet profits haven't yet been quantified in disclosed figures. Robotaxi's cumulative autonomous miles are nearly 2 million, against Waymo's more than 270 million miles through end of June 2026 (Waymo blog) — those benchmarks come from different points in time, and Waymo has operated much longer, so the comparison isn't perfectly apples-to-apples. Still, the scale difference is real. Even as the Robotaxi service expands to Dallas and Cybercab rolls out to more cities, assessing actual business scale is impossible when weekly trip counts, fleet sizes, and unit economics remain undisclosed. The rate at which that gap narrows is the defining long-term variable, and it won't be answerable until the company discloses the underlying operating data.
Optimus's external delivery timeline is the other long-range variable. Musk's "at least a year, maybe a little longer" before reaching 20,000 units a week and beginning external customer shipments means there's a meaningful runway even with Fremont production underway (per the Motley Fool transcript). Early Ramp status with no disclosed unit volumes or revenue is where you'd expect to be at this stage, but it also means the long-term bull case depends on proof points that haven't arrived yet. The departure of Model S/X production noted in the Form 10-Q as of mid-2026 is a product-mix shift worth tracking; what fills that revenue gap isn't clear from public filings. SpaceX-related transactions are worth monitoring: Tesla holds $2.0B invested in SpaceX and recorded $221M in Megapack revenue to SpaceX in Q3 — a related-party transaction structure that warrants ongoing governance attention given Musk's dual CEO role.
Let me be explicit about what changes my read. First, whether additional IEEPA tariff refunds are recognized in Q4: if they repeat, treating Q3's refund as a one-quarter variable needs revision; if they don't appear, Q3 becomes a distinctly unusual quarter in hindsight. Second, whether R&D excluding SBC — which I estimate rose roughly 74% year-over-year — produces tangible milestones in the filings. Third, whether Tesla begins disclosing software recurring revenue as a separate line; that single change would transform the quality of any transition analysis. Fourth, when Cybercab fleet count and weekly trip volumes first appear in public disclosures — that's the moment actual Robotaxi business scale becomes assessable. Fifth, the first external Optimus delivery, which marks the transition from manufacturing ramp to an actual commercial product sale. I see the third condition — software revenue disclosure — as the one most likely to shift my analytical baseline in the near term.
Here's where I could be wrong. If tariff refunds recur structurally in future quarters, my refund-adjusted operating income math overstates the underlying weakness. If FSD subscription growth sustains at this pace with low churn, the software revenue embedded in automotive sales may be building faster than I currently see. If energy storage deployments keep growing without needing refund pass-throughs to stay margin-positive, the energy segment's underlying health is stronger than the headline implies. If Optimus ships externally ahead of schedule or Robotaxi demonstrates profitability across multiple cities before Musk's estimate, the current CapEx load becomes rational frontloading in retrospect. I don't dismiss those scenarios — they're exactly what I'm tracking to decide whether my current read on the transition timeline is too cautious. But the public data in this filing doesn't yet provide enough evidence to confirm any of them. As the disclosures accumulate, the answers will become clearer, and so will whether my analysis was right or wrong.
This article is for informational purposes only and does not constitute a recommendation to buy or sell any securities. Stock investing involves the risk of loss of principal. Any investment decision should be made based on individual circumstances and after consulting a qualified financial professional.
Sources / References
- Tesla 10-Q (Q3 2026) — SEC EDGAR
- Tesla Q3 2026 Update (EX-99.1) — SEC EDGAR
- Tesla 10-Q (Q2 2026) — SEC EDGAR
- Tesla 8-K Credit Facility — SEC EDGAR
- Tesla DEF 14A — SEC EDGAR
- Ford 10-Q (Q3 2026) — SEC EDGAR
- Tesla Q3 2026 Earnings Call Transcript — The Motley Fool / Webull
- Automakers' Tariff Refund Treatment Varies in Quarterly Updates — Bloomberg Tax
- EY Global Tax News: IEEPA Ruling — EY
- Waymo Safety Data September 2026 — Waymo
- TSLA Historical Prices — StockAnalysis
- Tesla Q3 2026 Earnings Analysis — Money365.Market