In the stock market, there are two fundamental types of companies. The first type constitutes classical businesses, which are appraised in the antiquated manner: by their cash flows, by real revenue, and by marginality. This is mathematics. The second type is the so-called “dream companies.” Their valuation is built not on a calculator, but on pure faith that in the future they will be able to earn a lot of money.
Tesla (TSLA), unfortunately or fortunately, is not only a classical automotive company. With a price-earnings ratio that already exceeds the insane mark of 250x, and capitalization that eclipses all global automakers combined, the market tells us with extreme clarity: By purchasing Tesla shares, investors are not buying electric vehicles. They do not give a damn about cars. They are buying a lottery ticket to the world of robots.
For a long time, Optimus, Tesla’s humanoid robot, was simply a cherry on top, a pretty marketing bonus to the core business. A plaything for presentations. But the recent events of January 2026 altered everything.
Elon Musk, in his characteristic radical manner, made Tesla’s bet on robotics the main strategy of the company. I, as an analyst, see in this not only hype, but rigid engineering constraints. And I see a colossal risk. Between a beautiful presentation on stage and a mass, working product, there stands not merely complex code which can be finished over the weekend. No. A concrete wall of physical laws stands there. A wall that cannot be broken by the charisma of a CEO, even if his name is Elon Musk.
In this article I will analyze why the bet on 1 million robots by 2027 is a dangerous gambling game with physics, in which Tesla currently has no winning cards.
Before we plunge together into dreams about robots, let us look at the dry, boring numbers. Because any business boils down to one primitive ability — the ability to generate a stable cash flow.
And if we look attentively at the financial reporting of Tesla for the last fiscal period, we will see a picture that is located not simply in dissonance, but in a flagrant, screaming contradiction with its current market valuation. The capitalization of the company has reached the astronomical mark of $1.5 trillion dollars. This is more than the aggregate value of the giants Toyota (TM), Volkswagen (VWAGY), Ford (F), and General Motors (GM) combined.
But is this mountain of money supported by real, physical business growth? Fundamental data says “no.” The company’s revenue shows negative growth, or rather, a fall of 3.80% over the last year, having decreased from $97.7 billion to $94.8 billion.
Operating profit (EBIT) collapsed altogether, crashing by 38.1%, and earnings per share (without NRI) fell by 31%. This makes Tesla a mature, tired business, which has collided with a rigid cyclical downturn and saturation of demand.
Traditional auto giants, such as Toyota or Volkswagen, with similar or even larger volumes of real revenue, trade with a P/E multiplier in the region of 10x-15x. The market values them adequately. Tesla, however, trades with multiples characteristic of tech startups that double revenue annually. But the nuance is that instead of doubling, we see compression. The gap between the share price on the board and the real indicators of the business has become so colossal, so indecent, that it is already physically impossible to justify it by the sale of automobiles.
The current price is generally not a valuation of a business for the production of transport vehicles. It is a pure bet on a miracle. It is a religion. The market, in essence, tells us “we believe that robots will save everything and recoup any losses.” And it is precisely in this fanatical faith that the main, deadly risk for capital lies.
The events of recent weeks should have served as a cold shower. But the market, intoxicated by faith, perceived them as a signal for acquisition.
During its fourth quarter report, Tesla announced that it would cease producing the Model S and Model X, the automobiles that once made the brand a legend. The official reason? “Liberation of capacities at the factory in Fremont” for the production of 1 million Optimus robots per year.
The company, consciously, with its own hands, kills a line of premium products bringing live, real cash, to re-equip a working assembly line for a market that does not yet physically exist. This is a quintessential Musk move.
Investors sitting in soft armchairs see in this boldness and a visionary quality. I, however, see in this the desperation of an automotive business which has hit a concrete ceiling of growth and an attempt to escape into a new, even more complex niche. But the main question is: Is Optimus itself ready to become the “new iPhone?”
The principal, fatal delusion of Wall Street is contained in the following naive logic: “Since we created ChatGPT and it can write poems and code, it means that to create a smart robot is simply a question of time. Let us upload this AI into an iron body, and it is done.”
It is an error to think like this. Because for AI, there exists Moravec’s Paradox:
-
That which is complex for a human (playing chess, higher mathematics, writing poems) is incredibly easy for a computer.
-
That which is easy for a human (to walk across a room without tripping, to sit on a soft couch, to take a raw egg without crushing it) is incredibly difficult for a computer.
Modern large language models are, in essence, the cerebral cortex. They answer for logic, text, and meanings. But a robot in a workshop is not required to write poems. It needs a cerebellum — the part of the brain responsible for physical interaction with the chaos of the real world. For balance, for motor skills, for reflexes. And here we approach the main engineering problem, which is completely ignored by the current valuation of Tesla shares.
Let us conduct a simple thought experiment. Imagine that you bought Optimus so that it could help you around the house. You give the robot a simple, banal command: “Sit on the couch.” For you, this action is a reflex. For a robot, this constitutes a mathematical hell. To perform this action autonomously, the onboard computer of Optimus must perform a miracle:
-
Scan reality: Build a 3D map of the couch with accuracy to the millimeter. Comprehend where the soft cushion lies, and where the rigid wooden frame is concealed.
-
Calculate physics: This is not a solid floor. A couch is a soft, deformable body. To what extent will the foam rubber be compressed under the weight of 100 kg of metal? Where will the center of gravity be located in 0.5 seconds?
-
Process tactility: At the moment of contact, gigabytes of data regarding pressure will be transmitted by thousands of sensors on the “skin” of the robot. If the robot does not correct the tension of motors in 2-3 milliseconds (faster than an eye blinks), shaking will commence. And the robot will either fall or break your couch.
This requires computational power comparable to a modern server cabinet. And we run our nose into three physical barriers which make a mass-produced Optimus impossible in the stated terms.
Barrier A: Thermal Limit (Thermodynamics). To compute the physics of the real world in real time, chips of the NVIDIA H100 level or their future analogues are required. The consumption of such systems constitutes hundreds of watts, and with peripherals — kilowatts. A kilowatt of energy inevitably turns into a kilowatt of heat. Humans have a circulatory system for cooling the brain. The robot possesses only a plastic chassis. Tesla must figure out how to both install a computer capable of allowing a robot to sit on a couch into its head without melting the entire robot’s frame. Passive cooling here is insufficient, while active cooling (fans, liquid) requires space and energy, turning the robot into a noisy, hot vacuum cleaner.
Barrier B: Energy Density. The Tesla Model S has a huge flat battery on the bottom with a capacity of 100 kWh. Into the torso of a humanoid, you could likely physically place at most 2-3 kWh (otherwise he will be too heavy and will not be able to walk). If the “brain” of the robot consumes the energy necessary for full autonomy, plus powerful motors for walking, the autonomous operation time will constitute less than 1 hour. Who wants a robot assistant that functions for 40 minutes and subsequently charges for 3 hours? No one.
Barrier C: Silicon Limit. We cannot cram the power of a whole data center into the volume of a human skull.
Optimus, which will descend from the assembly line in 2027, will be the result of cruel, painful compromises. So that he does not overheat and works for at least 4 hours, Tesla engineers will have to radically, surgically “cut” his intellect. A lobotomy is inevitable.
-
He will have an excellent autopilot for walking on a flat concrete floor (Tesla knows how to do this).
-
He will be able to transport boxes from point A to point B in a straight line.
-
He will be able to dance at the presentation according to a pre-written script.
But the complex, fine, chaotic work which is expected of him cannot be performed. He will not be able to neatly load the dishwasher. He will not be able to change bed linens. He will not be capable of working on a construction site with unpredictable relief. And here lies the main financial trap. The market values Tesla in trillions of dollars. But if the product turns out to be a “robot with limited capabilities,” its utility instantly collapses to the market of warehouse logistics and expensive toys for geeks.
The difference in valuation between these two scenarios, I think, is huge for the current stock price. Imagine that you stood in line for an iPhone, and they sold you a very expensive pager. It works, it is technological, but it does not change the world. I think that when, say, 100,000 robots descend from the conveyor, the disappointment of consumers and investors will be devastating.
The fundamental architectural problem of the current Tesla strategy consists in the fact that it sets a bar that is too high.
Before humanity arrived at the Mercedes S-class, it was forced to travel a long, torturous path through three-wheeled carriages, through unreliable mechanisms that broke every kilometer, through decades of gradual, step-by-step hardware evolution.
In robotics, inevitably, the exact same evolutionary principle will work: from simple to complex. The history of technology teaches us that mass robotics will not start with humanoids. Instead, it will begin with specialized, utilitarian robots. For the execution of 90% of useful, economically profitable tasks, a robot does not need the form of a human.
-
To effectively clean the floor, an android with legs that can trip is not needed. A flat, stable “disk” (robot vacuum) is needed.
-
To carry weights at an Amazon warehouse, an expensive and fragile Optimus is not needed. A simple, self-propelled cart or “barrel” on wheels with a maximally low center of gravity is needed.
-
To perform complex repair works in pipes or dangerous structures, a robot-spider or flexible snake-like mechanism is needed, not a copy of a human.
Specialized robots are simpler in production, they are more reliable, and their brains require much less computational power energy. They are also much more effective in their specific, narrow niche. The path to a true universal android will take many decades.
I am absolutely convinced that robots will indeed go into series en masse, this is inevitable, but these will not be humanoids. These will be functional machines, sharpened for a specific task. Elon Musk, however, sells investors a beautiful idea, an illusion that we can “teleport” immediately to the finale of this evolution.
We observe a classic, textbook gap between the “Timing of the Visionary” and the “Timing of the Engineer.” Elon Musk is a visionary. He sees a world where robots do all the dirty work. And he is right — this future will inevitably arrive. But harsh engineering reality tells us that the hardware for this future is not ready yet. What does this mean for Tesla shares?
-
No Margin of Safety: The current stock price lays in an ideal scenario: a million robots are sold in 2027 and they change the world economy. Any delay, any technical problems (and they are inevitable due to physics) will lead to revaluation.
-
Cash Burn: Refusal of Model S/X deprives the company of a stable flow of revenue exactly at that moment when it needs billions for R&D of robots. This sharply increases the risk of cash gaps.
-
Speculative Hype: In the next half-year or year, we will see a mass of beautiful demo videos. Shares can grow on expectations and faith. But the moment of truth will arrive when the first real buyers receive robots and understand they cannot “sit on the couch.”
My recommendation: If you seek fundamental value, stay on the sidelines. Tesla now is not an investment in a business, it is a venture bet on the fact that Musk’s engineers will be able to violate the laws of thermodynamics in 2 years. I do not recommend opening short positions as the irrational enthusiasm of the market can last long. But I recommend fixing profit and reducing long positions.
On the date of publication, Mikhail Fedorov did not have (either directly or indirectly) positions in any of the securities mentioned in this article. All information and data in this article is solely for informational purposes. This article was originally published on Barchart.com