ZeroHero

Picture is worth a thousand words.

You mention FSD v12 here. Have you sat behind the wheel of FSD v14.2+ yet? As a FSD user since v10, the difference is night and day. The demos that I’ve seen online from are all fairly straightforward driving scenarios and possibly cherry picked. Assuming they aren’t cherry picked, There is nothing comparable to the hundreds of hours of full drives in various geographies that’s uploaded by Tesla’s customers. Even want-to-be content creators with 200 views per vid have uploaded hours of full drives.

If you want to know who has a chance to compete with FSD, apart from the nasty quantization tail that can only be solved with Hardware In the Loop for training, you should also look at the compute. Elon mentioned that you need one H100 to simulate one HD camera at real-time speed. That translates to eight H100s for the complete system. Simulation is necessary for Reinforcement Learning (RL), one of the most potent methods for training a neural network. Tesla has roughly 100,000 H100s available in Cortex. This is enough to simulate 12,500 full eight-camera systems in parallel. This is the best-case scenario, excluding gradient descent and all other overhead. Assuming the average driving speed of simulated scenarios is 50 km/h, you can simulate about 100 million km per week. However, you need to run each scenario several times — once per gradient descent iteration (or epoch). So if you assume 100 epochs (a reasonable ballpark guesstimate), you are down to 1 million km of RL training per week. If you want to reach superhuman safety, you need to train over several million kilometers. Considering you will need many iterations to add new scenarios, improve the reward function, try different network architectures, etc., you cannot get this done within a reasonable time if you do not have this amount of compute. Only Tesla has it. Nobody else is close. PS: To avoid any misunderstanding, I want to stress that I do not mean that millions of kilometers of data capture are enough to develop FSD. You need billions of miles captured by your fleet to discover the critical scenarios needed to train FSD (I explained that in a previous article).

Also, doesn’t appear that XPEV has that many EVs out on the road at all capable of capturing data intelligently. Key take away below: XPENG has not publicly disclosed exact numbers or percentages of vehicles equipped with the Turing chip. It is not standard across the entire lineup—even in newer models, it often appears in higher-end trims (e.g., Ultra/Max+ variants) rather than base models. • The equipped vehicles are concentrated in H2 2025 deliveries onward, particularly higher-spec trims of the G7 and subsequent launches. • Given that Turing rollout accelerated in late 2025 and into 2026, the number is in the tens of thousands at most (likely a minority share of the post-mid-2025 volume). The 2026 target of 550k–600k deliveries, combined with broader Turing deployment plans, will increase this share significantly over time. grok.com/share/bGVnYWN5…

@alojohhardcore @nymbusjp Eg this one, and many others that require the proper reactions from the driver that are next to impossible to source from data captures?

This is how every republic -> dictatorship transition starts. An ineffective government that no longer serves the people causes the people to turn towards singleton aggregate of strength.

@alojohhardcore Need a reset button for after we play with the toggles.

@elonmusk Nothing quite like back to back total world wars to accelerate technology.

Incredible that it was only 66 years from the first controlled, powered flight to landing on the Moon!

@alojohhardcore @kabans977 LMFAO at 2. That's hardcore af. Reminds me of Elon's deleting parts process

How do you show humans you have achieved AGI? How do you convince humans that the AI understands? Grok Imagine's output will be the acid test

Signal.

@HarmlessYardDog Generations of Mckinsey MBAs.

Which one of you bozos thought it was a good idea to build the entire world economy off just in time supply chains from the most unstable regions on the planet?

@bradsferguson Why doesn't the efficient markets know about it?

What if I told you there was a company trading at 5.5x next quarter's annualized profits, 0.06 price-to-earnings growth and forecast to more than double earnings... just had a better quarter than Tesla ever had... and profits will likely more than double from here?

@financialjuice I've never seen an organization be paid so much money and given so much attention to say "We don't know".

Fed's Powell: We just don't know what effects the rise in energy will be.

What's the point of paying all these thousands of PHDs at the Fed if all they ever say is "We don't know" ?

@elonmusk Time to build your own device!

iOS App Review delays are getting ridiculous