Editor’s Note: There’s a reason I spend so much time looking for what the market is telling us instead of simply accepting the story Wall Street is selling.
The best investors I know do the same thing — even if they use different tools to get there.
That’s one reason I’m excited to welcome Louis Navellier as my first ever live guest on Masters in Trading LIVE this Tuesday, Sept. 8, at 11 a.m. ET.[CW1]
Louis has spent more than four decades doing something that sounds simple but is incredibly difficult: following the numbers. He was an early pioneer of quantitative investing, building his approach around earnings growth, fundamentals and institutional buying pressure rather than whatever story happens to be captivating Wall Street that week.
We’ll have plenty to talk about Tuesday. I’ve been focused on copper as a potential choke point in the AI infrastructure buildout. Louis sees another bottleneck developing in memory and DRAM, one he believes could stretch well into 2027. We’ll compare notes on that, dig into his energy bets, and break down the jobs report, the Fed and what’s really happening underneath the S&P 500’s earnings growth.
Because you have to look under the headlines if you want to separate a great story from a great investment.
Few stocks have a bigger story attached to them right now than Tesla: Optimus, robotaxis, Grok, SpaceX and trillion-dollar markets. The possibilities are extraordinary. But possibilities and execution are two very different things.
So let’s dig into what’s actually happening today — and see what the numbers tell us about the bet TSLA investors are making.
On September 3 in Austin, Tesla took the wraps off the Cybercab — a two-seat robotaxi with no steering wheel and no pedals, built to drive itself on Tesla’s onboard AI.
It’s a fitting image for where the company now sits: the artificial intelligence the tech world spent two years arguing about as a chatbot is becoming something physical — a car that drives you, and eventually a robot that works beside you.
What’s the assistant running the conversation inside the car? None other than xAI’s Grok, which as of this summer controls the cabin of millions of Teslas through voice commands.
At least, that’s the story the market is buying.
Tesla today is really two businesses under one ticker: a car company with thinning margins, and a leveraged bet on its “physical AI” — robotaxis, the Optimus humanoid robot, and the Grok stack that increasingly runs both.
The bull case says Tesla is the only slice of Elon Musk’s sprawling AI empire you can own in a brokerage account.
The bear case says the dream is years away from reality and the car business funding it is under significant pressure.
In this piece, we’re going to examine exactly what separates the two: what Tesla has already shipped, what remains a promise, and what the company’s own filings reveal once the excitement of reveal day fades.
Grok is now driving the cabin — the “brain and body” thesis
The bull case for Tesla as an AI company rests on a simple split: Optimus and Full Self-Driving (FSD) are the body; Grok, from xAI, is the brain.
In 2026 those two halves physically connected.
Tesla’s 2026 Summer Update turned Grok from a novelty chatbot into a genuine vehicle controller.
Say “Hey Grok” hands-free and it now adjusts climate, makes calls, plays music, opens the glovebox, folds mirrors, toggles heated seats, and digs through settings buried in menus — chaining multiple actions from one plain-English sentence, running on xAI’s Grok Voice Think Fast 2.0.
Two caveats are worth noting. First, the setup requires AMD’s Ryzen infotainment chip, meaning older Intel-equipped cars are permanently excluded. Second, one community test found that Grok successfully handled 116 of 170 commands — a 68% success rate. Useful, but far from flawless.
As of late August, the fully integrated FSD-plus-Grok build has reached less than 1% of Tesla’s fleet.
The strategic advantage is distribution. OpenAI has an app. Anthropic has an app. Tesla has millions of machines sitting in driveways that just became Grok endpoints — with effectively no customer-acquisition cost because Musk’s companies control both sides of the equation.
The next step Tesla has signaled is putting Grok on top of FSD as a conversational “supervisor.” Tell the car, “Drop me at the entrance, then go park,” and Grok would translate that request into actions for FSD. That’s exactly the kind of interface a driverless robotaxi passenger would need. But for now, treat it as a roadmap feature, not something Tesla has shipped.
Now let’s dig into the ecosystem TSLA is plugged into. Buckle up, because things are about to get wild.
To understand where Tesla fits into the AI race, it helps to zoom out. Because the network forming around the company is starting to look less like a collection of Musk ventures and more like an interconnected technology stack.
Musk’s xAI merged with SpaceX in early 2026. SpaceX then went public in June in the largest IPO on record. From there, the combined SpaceX/xAI agreed to acquire Cursor-maker Anysphere for $60 billion in stock — the largest acquisition of a venture-backed startup ever.
But perhaps the strangest connection comes from a rival. Anthropic agreed to pay xAI roughly $1.25 billion every month through May 2029 — more than $40 billion in total — for access to xAI’s Colossus 1 supercomputer in Memphis. In other words, xAI built so much computing capacity that, when Grok usage fell short of it, one of its biggest AI competitors stepped in to rent the excess.
Tesla isn’t directly signing these deals. But it doesn’t sit outside this ecosystem, either. It owns a stake in SpaceX, Grok is moving deeper into Tesla vehicles, and the boundaries between Musk’s companies are becoming increasingly blurred. That means Tesla is plugged into a network spanning cars, rockets, satellites, AI models, coding software and some of the largest computing infrastructure on Earth.
Here’s a fact almost no casual TSLA holder knows they’re exposed to — and it’s the bridge to the entire AI story.
Tesla’s most recent quarter reported $1.11billion of GAAP net income. Roughly $ 1 billion of that was an unrealized gain on Tesla’s equity stake in SpaceX — a paper mark, not operating cash. Strip it out (as Tesla itself does in its non-GAAP figures) and the quarter’s “profit” looks far thinner.
So where did that stake come from? Tesla originally invested about $2 billion in xAI. When xAI merged into SpaceX, Tesla received U.S. regulatory clearance in March 2026 to convert that investment into roughly 19 million shares of SpaceX Class A stock — giving Tesla a stake of less than 1%.
Then came the IPO. SpaceX went public on June 12, 2026, and its valuation surged past $2 trillion within days, pushing the value of Tesla’s stake well above its original $2 billion investment
That matters for more than bragging rights. Tesla carries the investment at fair value, meaning it revalues the stake every quarter. As SpaceX shares rise or fall, those changes can flow through Tesla’s reported earnings — potentially giving the bottom line a significant boost or drag from one quarter to the next.
And there’s one more date worth watching: December 2026, when Tesla’s restrictions on selling those shares expire.Why this matters for a retail trader: when you buy TSLA, you are also buying a slice of SpaceX — and therefore of xAI and Grok. That exposure is genuine, but it’s also a source of earnings noise. A chunk of a future “beat” or “miss” could be nothing more than SpaceX’s mark moving around. Learn to read past it.
- Where it’s built: Fremont, California — the former site for the Model S and X lines, now decommissioned and being rebuilt for Optimus. A second, far larger factory is under construction at Giga Texas.
- Capacity ambition: Up to 1 million units per year at Fremont and 10 million per year at Giga Texas over the long term.
- Target price (long-term, at scale): Roughly 30,000 — Musk’s aspiration, not a near-term sticker.
- The brain: A vision-based neural-network stack adapted from Tesla’s FSD software, with Grok handling natural language on newer versions.
- How it learns: By imitating humans on first-person video, refining in simulation, and sharing skills across the fleet.
- External sales: Targeted “as early as” the second half of 2027, per a JPMorgan note after a Fremont visit.
Treat online Gen 3 “spec sheets” with caution — Tesla has not officially revealed the robot, and most recycle older figures.
This is where careful investors separate the milestone from the marketing.
What’s confirmed: Tesla has decommissioned the Model S and Model X lines at Fremont and is installing its first Optimus production lines in their place — a transition documented with photos in the company’s Q2 2026 update. Pilot and data-collection units have also been operating since early 2026, performing narrow tasks such as sorting battery cells. Tesla’s installed-capacity table lists Optimus facilities in both California and Texas, but both remain classified as “Construction,” with no production capacity disclosed.
What isn’t confirmed: Full-scale production or external sales. Tesla still describes Optimus production on the new line as anticipated, without providing a firm start date, production target or price. And the first units off the line are expected to remain inside Tesla for training and data collection rather than go to customers.
That distinction matters because Tesla has been here before. Musk projected production of roughly 10,000 Optimus units in 2025. Tesla ultimately built only a small fraction of that number, and Musk acknowledged in early 2026 that none were yet performing “useful work” in its factories.
None of that makes the Fremont conversion insignificant. Retiring a vehicle line and replacing it with robot production is a real, expensive commitment — and much harder to dismiss than a prototype on a stage. But it’s also important not to confuse commitment with execution. Starting production and producing at meaningful scale are two very different milestones. Musk himself has described the Optimus ramp as “agonizingly slow.”
The optimists aren’t arguing about next quarter. Instead, they’re arguing about the size of the prize.
Venture capitalist Jason Calacanis represents the extreme end of that optimism. He has called Optimus “the greatest product ever made by humanity” and predicted that humanoid robots could reach 1 billion units worldwide by 2036. That’s a prediction, not a forecast grounded in anything close to today’s production. But it captures the sheer scale of the opportunity Tesla bulls believe they’re underwriting.
There’s a more sober way to make the case: follow the economics of labor. A Citizens Bank analysis, highlighted by Futurum strategist Shay Boloor, estimates Optimus could eventually address a roughly $1.7 trillion U.S. market, with about $300 billion potentially serviceable in the nearer term across factories, warehouses and back-of-house jobs. The basic math is compelling: The analysis models a humanoid robot costing roughly $5 an hour to operate, compared with about $35 an hour for a human worker.
And this is where Tesla may have an advantage that most robotics startups simply don’t. It can be its own first customer. Tesla can deploy Optimus inside its factories, use real-world work to improve the robots and potentially lower its own labor costs — all while proving the technology before selling a single unit to anyone else.
Combine that with the Grok distribution advantage we discussed above, and you get the full bull thesis: Tesla gives public-market investors a way to own a piece of Musk’s increasingly interconnected AI stack — from intelligence and autonomy to the machines that could eventually put both to work in the physical world.
For every bull number, there’s a counterweight — and lately the counterweights own the near term.
Prediction markets have turned cautious: on Polymarket, the odds of a commercial Optimus launch by end-2026 recently sat around 9%, down from a peak near 33% a month earlier. Musk himself calls Optimus Tesla’s hardest product to scale, because it requires an entirely new supply chain, and the Gen 3 reveal has slipped repeatedly without the robot being shown. The skepticism isn’t limited to bears: Tesla supporter Ross Gerber has questioned Optimus’s near-term commercial prospects, and Unitree founder Wang Xingxing has suggested humanoids may be years from their “ChatGPT moment.”
There’s also regulatory risk that grows as Grok moves deeper into the car — and as the Cybercab hits the road. The September 3 Austin event was a product unveiling, not a regulatory green light: the two-seat, wheel-free Cybercab still faces limited public access, and Tesla’s autonomy programs already sit under open federal investigations. An AI that operates vehicle functions — and eventually a driverless car with no human fallback — invites scrutiny. Expect headlines on that front.
Strip away the reveal-day buzz and the billion-robot predictions, and here’s the last hard financial data we have on the business paying for all of it. Tesla’s most recent report — Q2 2026, for the quarter ended June 30 — tells a sobering story beneath the record headline.
Metric (Q2 2026)FigureYear-over-yearTotal revenue$28.24B+26% (record; first-ever $100B+ trailing-twelve-month)Automotive revenue$20.52B+23%— Regulatory credits$146M
−67% (from $439M)Energy generation & storage$3.14B+13%Services & other$4.58B+50% (record margin)Total gross margin~16.8%Auto GM 16.9% (16.3% ex-credits)
Operating income$398M−57%Operating margin1.4%down from 4.1%Operating expenses$4.35B+47% (AI, Optimus, robotaxi, SBC)GAAP net income$1.11B−5% (includes ~$1.0B SpaceX gain)Non-GAAP EPS$0.33missed ~$0.53 est.; −18%Operating cash flow$4.70B+85%Capital expenditures$5.79B
+142%Free cash flow−$1.09Bfirst cash-burning quarter since early 2024Cash + investments$43.52B+18%Deliveries480,126record for the quarter
Record revenue, vanishing profit. Revenue hit an all-time high and Tesla crossed $100 billion in trailing-twelve-month sales for the first time. Yet operating income fell 57% to $398 million, compressing operating margin to just 1.4%. The top line grew; almost none of it reached the bottom line. Operating expenses jumped 47% as Tesla poured money into AI, Optimus, and robotaxi, plus stock-based compensation tied to the 2025 CEO pay package.
The regulatory-credit cliff is real. Automotive regulatory credits — nearly free profit Tesla has banked for years — fell 67% to $146 million from $439 million, as governmental actions restricted the programs tied to Tesla’s products. Credit revenue peaked at a record $2.76 billion in 2024. That cushion is deflating fast, and it flatters no future quarter.
The company is now burning cash. Operating cash flow rose 85% to $4.7 billion, but capital spending more than doubled to $5.79 billion, pushing free cash flow to negative $1.09 billion — Tesla’s first cash-burning quarter since early 2024. Management guided full-year 2026 capex above $25 billion and is arranging up to roughly $30 billion in borrowing capacity. Tesla ended the quarter with a still-healthy $43.5 billion in cash and investments against about $9.1 billion in principal debt, so this is an investment cycle, not a liquidity scare — but the “self-funding growth machine” narrative is on pause.
For a retail trader, the one-line takeaway is this: Tesla’s fundamentals are being re-rated on quality of earnings. The business is spending aggressively today for products that don’t generate revenue yet. Whether you’re bullish or bearish, that’s the lens.
The longer-term signal is in Austin. Permit-level filings point to two major new Giga Texas buildings: a dedicated Optimus factory and an Advanced Technology Chip Fabrication facility — the latter reportedly a joint Tesla–SpaceX effort of roughly 470,000–490,000 square feet with a specialized vibration-mitigation foundation typical of precision chipmaking. Read together, the plan is tighter vertical integration: build the robots and the silicon that runs them on one campus. But it’s years and billions from revenue, with meaningful Texas Optimus output not expected until around summer 2027.
Put the pieces together and the investable summary is straightforward:
- The car business funds the bet, and it’s under pressure. Shrinking margins, collapsing regulatory credits, and negative free cash flow mean the “physical AI” ambitions are being financed through a heavier-spending, thinner-earning core than a year ago.
- The AI upside is real but priced-to-perfection. Robotaxi, Optimus, and the Grok stack are genuine option value — and the market is clearly paying for them, with TSLA carrying a roughly $1.4 trillion market cap despite the earnings compression, down more than 20% year-to-date even after an ~18% August rally.
- Watch the quality of earnings, not the headline. A future “beat” driven by the SpaceX mark or a bitcoin gain is not the same as operating strength. Separate the paper gains from the cash the business actually produces.
- It’s closer to binary than most stocks. If Optimus and autonomy land, Tesla is early in enormous new categories. If they stumble, it has poured capital into hard-to-reverse capacity against already-thin margins.
None of that is a call to buy or sell. It’s the framework for making your own call with your eyes open.
The September 3 Cybercab launch — fleet size and whether riders can actually hail one in the days after is the first real read on how fast the robotaxi bet scales.
The Optimus Gen 3 reveal — Tesla says it’s withholding the design until production is close; an actual reveal is the first hard timeline signal.
A disclosed Optimus production number — it has never appeared in Tesla’s production reports.
Q3 2026 earnings — the next read on margins, regulatory credits, and free cash flow, and whether Optimus finally shows up as a line item.
The SpaceX share lockup expiring December 2026 — a date that could affect how Tesla’s stake and reported earnings behave.
Grok-on-FSD rollout — the conversational “supervisor” layer landing in the fleet would validate the brain-and-body thesis.
Yes. Tesla holds roughly 19 million SpaceX Class A shares — under 1% of the company — after converting an earlier ~$2 billion investment in xAI into SpaceX stock when the two merged. Tesla recorded about $1 billion in unrealized gains on that stake in the first half of 2026, and sales restrictions on the shares expire in December 2026.
In its most recent quarter, revenue rose 26% to a record $28.24 billion, but operating income dropped 57% to $398 million. Operating expenses jumped 47% on AI, Optimus, and robotaxi spending, and high-margin regulatory-credit revenue fell 67% — so almost none of the extra revenue reached the bottom line.
Not full production. Tesla has decommissioned its Fremont Model S/X lines and is installing Optimus lines, with pilot and data-collection units running. But the company still describes production as “anticipated,” with no official start date, unit count, or price, and early units go to internal training rather than customers.
Tesla has targeted external commercial sales “as early as” the second half of 2027, with a long-term at-scale price target of roughly 30,000. That price is an aspiration, not a confirmed near-term figure — and given repeated delays, the timeline should be treated as tentative.
The Cybercab is Tesla’s purpose-built two-seat robotaxi with no steering wheel and no pedals, designed to run entirely on Tesla’s Full Self-Driving software. Its formal launch event was held September 3, 2026, in Austin — a product unveiling and limited deployment, not full public availability.
After the 2026 Summer Update, xAI’s Grok can control vehicle functions by voice — climate, calls, music, glovebox, seats, mirrors, and settings search — hands-free via “Hey Grok.” It requires the AMD Ryzen infotainment processor, so older Intel-based cars are excluded.
Tesla trades around a $1.4 trillion market cap despite compressed earnings, so the market is clearly assigning meaningful value to Optimus, robotaxi, and AI. That makes execution — a Gen 3 reveal, disclosed production numbers, and robots doing paid work — the key question for whether the valuation holds up.
The Cybercab reveal is the visible version of Tesla’s whole thesis: AI you can sit inside. But the durable story is quieter and harder. Tesla’s own filings show a record-revenue business with thin, credit-dependent profits, now burning cash to fund an audacious bet — a bottom line flattered by a billion-dollar paper gain on a SpaceX stake that quietly links TSLA to the entire Grok/xAI universe. Layered on top is the biggest and most speculative wager of all: Optimus, a robot Tesla hasn’t officially shown, on sale to outsiders no sooner than late 2027, with single-digit near-term launch odds and trillion-dollar dreams attached.
For retail traders, the discipline is the same in either direction: separate the cash from the marks, and watch execution instead of adjectives. Production counts, a real reveal, and robots doing paid work will tell you more than any billion-unit prediction. Until then, TSLA is a story stock inside a story stock — enormous if it lands, expensive if it doesn’t.
This article is for informational and educational purposes only and is not investment advice or a recommendation to buy or sell any security. Financial figures are drawn from Tesla’s Q2 2026 shareholder update and Form 10-Q for the quarter ended June 30, 2026, and other cited public sources; they are historical and will change with future filings. Statements about Optimus, Grok, the Cybercab, capacity, pricing, and timelines are company targets, analyst estimates, or third-party predictions that are forward-looking and unconfirmed — Tesla has not officially revealed the Gen 3 robot or announced an Optimus commercial launch date. Stock prices and market data are as of September 3, 2026. Do your own research and consider consulting a licensed financial professional before making investment decisions.
The post Tesla Optimus, Grok, and TSLA Stock: Inside the Physical-AI Bet That Now Defines Tesla appeared first on InvestorPlace.
