Tesla’s VP of AI/Autopilot software, Ashok Elluswamy, has offered a rare inside look at how the company’s AI system learns to drive. After speaking at the International Conference on Computer Vision, Elluswamy shared details of Tesla’s “end-to-end” neural network in a post on social media platform X.

How Tesla’s end-to-end system differs from competitors

As per Elluswamy’s post, most other autonomous driving companies rely on modular, sensor-heavy systems that separate perception, planning, and control. In contrast, Tesla’s approach, the VP stated, links all of these together into one continuously trained neural network. “The gradients flow all the way from controls to sensor inputs, thus optimizing the entire network holistically,” he explained.

He noted that the benefit of this architecture is scalability and alignment with human-like reasoning. Using examples from real-world driving, he said Tesla’s AI can learn subtle value judgments, such as deciding whether to drive around a puddle or briefly enter an empty oncoming lane. “Self-driving cars are constantly subject to mini-trolley problems,” Elluswamy wrote. “By training on human data, the robots learn values that are aligned with what humans value.”

This system, Elluswamy stressed, allows the AI to interpret nuanced intent, such as whether animals on the road intend to cross or stay put. These nuances are quite difficult to code manually.

Tackling scale, interpretability, and simulation

Elluswamy acknowledged that the challenges are immense. Tesla’s AI processes billions of “input tokens” from multiple cameras, navigation maps, and kinematic data. To handle that scale, the company’s global fleet provides what he called a “Niagara Falls of data,” generating the equivalent of 500 years of driving every day. Sophisticated data pipelines then curate the most valuable training samples.

Tesla built tools to make its network interpretable and testable. The company’s Generative Gaussian Splatting method can reconstruct 3D scenes in milliseconds and model dynamic objects without complex setup. Apart from this, Tesla’s neural world simulator allows engineers to safely test new driving models in realistic virtual environments, generating high-resolution, causal responses in real time.

Elluswamy concluded that this same architecture will eventually extend to Optimus, Tesla’s humanoid robot. “The work done here will tremendously benefit all of humanity,” he said, calling Tesla “the best place to work on AI on the planet currently.”

Tesla is releasing a modified version of the Full Self-Driving (Supervised) version 14 suite for Hardware 3 owners, the company announced during the Q3 Earnings Call earlier this week.

Perhaps one of the most pertinent issues for Tesla owners right now is that those owners who have purchased vehicles before 2024 have been stuck with an older version of the company’s self-driving chip, known as Hardware 3 or AI3.

Owners with Hardware 3 vehicles have been stuck in a strange limbo for some time, wondering whether they should wait for the company’s official plans to upgrade them to the newer AI4 or even the to-be-released AI5 chip, or if they should purchase a new vehicle altogether. The upgrade would give them access to the latest Full Self-Driving suite releases, but it would likely cost a good bit of money.

Tesla (TSLA) Q3 2025 earnings: Wall Street’s reactions

For a while, these owners have been waiting for Tesla to give some sort of update on its plans, as the company has, in a way, danced around the issue by stating it would “take care” of those owners. The problem is, the definition of “take care” is subjective, and nobody knows if that means an upgrade or a free Tesla t-shirt.

Nevertheless, many owners finally got a tad bit more color earlier this week during the Earnings Call, when company executives finally outlined the beginning of a concrete plan to “take care” of HW3 vehicles.

Chief Financial Officer Vaibhav Taneja gave the first bit of the answer, as it is a personal issue to him. He also said that the vehicle he drives is a HW3 car, so it is impacted by the lack of upgrades.

He said:

“We have not completely given up on HW3. These customers are very important. They are early adopters. We will definitely take care of you guys.”

However, Tesla’s Head of AI, Ashok Elluswamy, gave some additional color, revealing that Tesla plans to launch v14 Lite for HW3 cars, and it will be released in Q2 of next year, tentatively:

“Once the v14 release series is fully done, we are planning on working on a v14 Lite version for hardware three. Probably expected in Q2 next year.”

This is somewhat of an answer, but some owners have already voiced discontent with this solution because HW3 will more than likely not be capable of what will be the “feature complete” version of FSD.

Investor and host of Mad Money on MSNBC , Jim Cramer, has chimed in on Tesla CEO Elon Musk’s pay package and whether it should be rewarded to the frontman or not.

Cramer has drawn a lot of attention regarding his sentiments on Tesla, as investors have routinely given him a pretty hard time over what he’s said about the company.

For the past few years, we have covered his comments on Tesla when he has something to say, mostly because his opinion on the stock seems to change pretty frequently; at a minimum, he has something different to say about it every few months.

However, Cramer knows Musk’s value to Tesla, and said on Thursday that he believes the CEO deserves his pay package:

“Don’t be small-minded: Tesla is about robots, Full Self-Driving, the future. Give him his package.”

Don’t be small-minded: tesla is about robots, full self driving, the future. Give him his package

— Jim Cramer (@jimcramer) October 23, 2025

Cramer’s comments come just one day after Tesla’s Q3 2025 Earnings Call, where Musk took several opportunities to call out the importance of the pay package and how it could impact the company’s future — with or without him.

Musk said at one point that he would not feel comfortable continuing to develop the company’s massive fleet of Optimus bots without having appropriate control of the company from a voting perspective.

He said he does not want so much power that if he “were to lose his mind,” he could not be removed. However, he does feel he needs to be protected from “activist shareholders,” or “corporate terrorists” like proxy groups Institutional Shareholder Services (ISS) and Glass Lewis:

“My fundamental concern with regard to how much voting control I have at Tesla is if I go ahead and build this enormous robot army, can I just be ousted at some point in the future? …It’s just, if we build this robot army, do I have at least a strong influence over that robot army, not current control, but a strong influence? That’s what it comes down to in a nutshell. I don’t feel comfortable wielding that robot army if I don’t have at least a strong influence.”

At the end of the call, Musk said:

“Like I said, I just don’t feel comfortable building a robot army here and then being ousted because of some asinine recommendations from ISS and Glass Lewis, who have no freaking clue. I mean, those guys are corporate terrorists.”

Cramer is one of many who realize Musk’s importance to Tesla, and how the company would likely lack the guidance and prowess it does without his planning and drive. However, Tesla shareholders will have the ultimate say on November 6 when they vote on Musk’s compensation plan.