This Could Be A Death Blow For Tesla
Musk admitted he lied to everyone.

With crashing sales, plummeting profits, vehicle launches that flopped so badly they’re openly mocked in the street, a glacially slow Robotaxi rollout, and the loss of its monopolistic hold on the EV market, Tesla is on a generational downward spiral. But you wouldn’t know that from its rhetoric, or even its share price. It’s almost like reality doesn’t matter. However, that might change soon, as Musk has just admitted something that could not only put the final nail in Tesla’s financial coffin but also destroy the myth that props up its insane value. You have probably seen this news plastered everywhere by now, but Musk has confessed that, despite his previous claims, HW3 can’t achieve full autonomy, and millions of customers who bought a car with the promise that it will one day drive itself will need to upgrade. The implications of such a damning admission go a lot deeper than you might think.
But, before we dig into everything, let’s quickly recap how the whole mess started.
The Hardware Debacle
Way back in 2017, Tesla proudly proclaimed that it sold all its vehicles with “all the hardware necessary for full self-driving capability”. Owners were promised that, at some point down the line, the only thing it would take is a software update, and their car would suddenly be able to drive itself — no human input required. In short, Musk promised that the HW2.5 computer and sensor suite installed in all sold Teslas would one day be capable of full autonomy.
But in 2019, that narrative began to fall apart. Tesla launched its updated HW3, which had a substantially more powerful computer. Why? Well, the self-driving AI Tesla had developed was simply too large to run on the HW2.5 computer. Musk and Tesla had flat-out lied about its capability. All new Teslas going forward would have HW3, and Musk rolled out a free upgrade program to anyone who bought a HW2.5 Tesla with the FSD (Full Self-Driving) package. But anyone who wanted to upgrade that hadn’t bought FSD was expected to shell out $1,500 (later discounted to $1,000).
Then, even worse, this entire debacle repeated itself. In early 2023, Tesla launched its HW4 computer, which again was dramatically more powerful than the previous generation, with eight times the memory bandwidth! It was widely known that HW3 simply didn’t have the power for full autonomy, and in January 2025, Musk finally admitted to it.
There were some rumours that a kind of ‘FSD lite’ software would be available for HW3 vehicles. But now, Musk has fully confirmed that it’s not happening. During a recent earnings call, Musk confessed that HW3 can’t cut the mustard. “Unfortunately, Hardware 3 — I wish it were otherwise — but Hardware 3 simply does not have the capability to achieve unsupervised FSD.” His solution is a rinse and repeat from last time. “I do think over time its gonna make sense for us to convert all HW3 cars to HW4, because that’s what enables them to enter the Robotaxi fleet and have unsupervised FSD,” Musk said before he proposed setting up ‘microfactories’ to convert Tesla’s fleet.
The Scale
But what Musk failed to mention is the sheer scale of such an operation, because Tesla has sold a lot more cars since 2019.
Electrek estimates that there are roughly four million Teslas with HW3. That is roughly eight times the number of Teslas that were fitted with HW2.5! An HW4 upgrade includes not just a much more capable computer, which in the current chip market will be substantially more expensive than the HW3 computer, but also upgrading the camera suite FSD uses. There will also be significant overheads for these ‘microfactories’ needed to process the sheer volume of upgrades.
Tesla sold those vehicles with the explicit promise that they would one day soon drive themselves. To avoid running foul of fraud laws, Tesla will either have to upgrade all these vehicles at its own expense (which Musk has hinted towards) or compensate these owners.
So, how much will that cost Tesla?
Well, let’s be incredibly generous and assume an HW4 upgrade will cost Tesla $2,000, or $500 more than the HW3 upgrade cost consumers. Multiply that by four million, and you get a total bill of $8 billion!
You might think Tesla is large enough to absorb such a cost without much issue, but it isn’t. Especially right now. Tesla’s total profit for 2025 was just $4.4 billion. Given the lowballing nature of my estimate, this bill could easily wipe out Tesla’s profits for two years straight.
This is assuming Tesla has to upgrade all these cars. In order to force Tesla’s hand, you could make a case for something like a class-action lawsuit. But these things are never so certain, particularly when it comes to someone as slimy as Musk.
However, Musk admitting HW3 is a failure and that Tesla will need to do some sort of upgrade scheme has overarching implications, many of which people don’t seem to be discussing at all. So, let’s break them down in no particular order.
Financial Implications
Tesla is expected to triple its capex this year, almost entirely driven by AI spending. This, combined with shrinking sales, means Tesla won’t be cash-flow positive for the rest of the year. In other words, Tesla will most likely operate at a loss this year. So announcing that Tesla will be forced to shell out billions of dollars to upgrade or compensate millions of customers because the CEO and company lied about a product’s capability is an enormous blow. It means this cost won’t gently tip them into the red — it will bury them in it.
Accountability Implications
On that note, if any other CEO had announced that a lie they had been heavily pushing for years was about to cost the company billions of dollars to correct, they would face severe disciplinary actions. We’re talking about more than simply being ousted. The board has the option to try to hold Musk personally liable, given that this lie originated solely with him. For example, they could sue him personally for the damages he has caused the business. If Tesla’s compliance and management were even vaguely functional, this would be the natural course of action. But no, nothing. Musk just admitted to blatantly lying to Tesla customers for a decade, and not just a small lie, but one that undermines something critical to the overall value of Tesla. Yet nothing has happened to him.
This demonstrates exactly how broken Tesla is. It proves it requires zero accountability, and that is a recipe for catastrophe.
Musk’s Capability Implications
But this also heavily implies that Musk is not capable.
By making a mistake this severe, it suggests he was either too stupid to understand the task at hand or he was lying and grifting the whole time. There is no other option here. Either he believed HW3 was capable — despite it demonstrably lacking the oomph needed to run the gigantic AI that Tesla’s moronic vision-only system requires, which completely writes him off as a competent leader — or he knew it wasn’t going to be good enough but openly lied about it anyway to grift and pump Tesla’s stock (which, again, disqualifies him as a competent leader).
Either way, this admission paints Musk as dangerously incompetent. Especially when it’s the second time he has done this! Despite his claims, both HW2.5 and HW3 were nowhere near good enough. Therefore, he is automatically a repeat moron or a repeat charlatan.
Future Implications
But who is to say that HW4 will solve the problem?
I, like many autonomous experts, believe that FSD fundamentally can’t reach safe autonomy. The vision-only system lacks any redundancy, meaning it requires nearly 100% accurate AI to function safely, and even then, only in ideal conditions. But AI physically can’t reach those levels of accuracy. So, it doesn’t matter how good HW4 is; there is a reasonable chance that the system will require a complete overhaul in the future, with a suite of new sensors like radar and lidar, and more powerful computers capable of running multiple AI models to cross-reference each other and create these layers of redundancy. If you want to know more about this topic, read my previous article here.
Keep in mind that Musk believed HW2.5 and HW3 were good enough. What are the chances that he has got it wrong again? Statistically speaking, it’s damn high.
But by far the most telling sign is that Tesla is already getting ready to launch HW5, also known as AI5. This computer is supposed to be ten times more powerful than HW4! I ask you, if Musk or Tesla had faith that HE4 was good enough, why would they be moving to replace it with something astronomically more powerful just a few years after launching the HW4?
There are currently around three million HW4 Teslas on the road, and Tesla is presently selling more than a million a year. So could Tesla be forced to upgrade another four or five million vehicles to AI5 in a few years’ time for another multi-billion-dollar price tag?
Possibly, and that should terrify anyone connected to Tesla.
Summary
This response skirts around the deeply broken ideology at Tesla. It is no secret that the Tesla board and investors value Musk’s lying. His ‘overexaggerations’ about Tesla’s capabilities are what have made the company so ludicrously valuable, after all. But sooner or later, reality will come home to roost. This has the potential to be the first time Tesla is pressured to sit up and pay attention. Rather than boosting its value, this lie is set to pile Tesla deeper and deeper into the red. As soon as they realise his falsehoods are a liability and not a boon, I wonder how quickly they will all turn on him.
Thanks for reading! Everything expressed in this article is my opinion, and should not be taken as financial advice or accusations. Don’t forget to check out my YouTubechannel for more from me, or Subscribe. Oh, and don’t forget to hit the share button below to get the word out!


Thanks to your articles, I am keeping track of the ways AI hype has failed in the real world. I have watched multiple interviews with AI experts warning about the threat of AI wiping out humanity or at least eliminating “90% of jobs!!” Never do the interviewers bring up the business application failures, the computing limits or in this case, the limited self-driving capabilities. If AI is going to destroy mankind, with or without gaining consciousness, it better hurry up, before users get fed up and drop it and/or the circular financing bubble pops and wipes out AI companies first. And then there’s the growing pushback against swamping rural communities with energy- and water-guzzling data centers. We need more exposés of the lies of Musk and Altman and the fascist fantasies of Theil, Yarvin and Karp.
Let me confirm that large companies have finally gotten from the adoption phase to the “please save money phase.” The cost of compute and tokens is more than the people who can do the work at the level at AI. It is also safer than AI.
Having used it for my portion of tech I can say it helps to point out, or correlate where there might be something useful. It is not a replacement for discernment. It takes time off of the clock. It does not replace.