When Nvidia Corp (NVDA) showed off its new autonomous driving technology at the Consumer Electronics Show this week, Tesla Inc. (TSLA) CEO Elon Musk had some thoughts. And they weren't dismissive, exactly. More like... knowingly sympathetic.
The Long Tail Problem
Responding to a post on X about Nvidia's announcement, Musk offered what sounds like hard-won wisdom. "What they will find is that it's easy to get to 99% and then super hard to solve the long tail of the distribution," he wrote. Translation: getting self-driving tech to work most of the time is achievable. Getting it to handle every weird edge case that pops up in the real world? That's where things get brutal.
But Musk wasn't throwing shade. "I honestly hope they succeed," he added, which feels surprisingly gracious for someone whose company is directly competing in this space.
Tesla's AI Chief Weighs In
Ashok Elluswamy, Tesla's head of AI, jumped into the conversation to back up his boss. "The long tail is sooo long, that most people can't grasp it," he posted. Elluswamy has previously highlighted Tesla's chip-building capabilities and the company's integrated approach to developing AI hardware, suggesting that vertical integration might be key to cracking this problem.
What Nvidia Actually Announced
So what has Nvidia cooking? At CES 2026, CEO Jensen Huang unveiled Alpamayo, a model built on Vision-Language-Action technology that supposedly brings human-like reasoning to autonomous systems. "The ChatGPT moment for physical AI is here — when machines begin to understand, reason and act in the real world," Huang declared. He specifically mentioned robotaxis as a beneficiary of the technology.
Nvidia isn't just building tech in a vacuum, either. The chipmaker announced a partnership with Mercedes-Benz to deliver a Level 2 Driver Assistance system powered by Nvidia's full-stack autonomous vehicle software. This follows similar collaborations with European automakers that kicked off last year.
Where Tesla Stands on Robotaxis
Meanwhile, Tesla continues pushing forward with its own robotaxi ambitions, though not quite at the pace Musk originally projected. The company missed its goal of launching driverless operations in Austin by the end of last year. Still, Musk shared that a Tesla Model Y robotaxi drove him around autonomously in late 2024, and that's not nothing.
Multiple Cybercab prototypes have been spotted testing across Austin and California, giving the public glimpses of what Tesla's purpose-built robotaxi might look like in action. Musk has said production of the vehicle will scale up during 2026, though anyone who's followed his timelines knows to take that with appropriate caution.
The Real Challenge
What makes Musk's comments interesting is that he's not really arguing about the technology itself. Nvidia has serious AI chops, and nobody's questioning that. The issue he's flagging is distribution, which in this context means deploying self-driving systems that work reliably across all the unpredictable scenarios real-world driving throws at you. That last few percentage points of reliability requires solving for thousands of rare but critical situations, and that's exponentially harder than the initial development.
It's a reminder that in autonomous driving, getting close isn't good enough. The technology needs to handle not just normal conditions, but also the weird stuff: unexpected road closures, unusual weather, construction zones configured in ways nobody anticipated, pedestrians doing unpredictable things. That's the long tail Musk and Elluswamy are talking about.
Price Action: Tesla (TSLA) shares declined 0.14% to $451.05 in after-hours trading.