The compute cost for DeepSeek V3 was $5.6M. Training costs for traditional US frontier labs model is about $70 to $200M+, with estimates of up to $1B+ to train the next generation models. It's expensive, but it's not even in the ballpark of "There is no viable economic model to do so." Google revenue in Q2 was $1.32 billion *per day*. $1B is trivially less than a day; $5.6M is just 6 minutes of revenue.

The cost problem is in the compute buildout and the subsidized compute to capture users.

> It remains to be seen if even using their models can be profitable,

I presume we can stipulate there exists some tasks that are cheaper to do using an LLM than doing it "by hand". As a such, using an LLM model can be profitable. To me, it is also obvious that the business of serving an LLM model can be profitable. There's large benefits to being able to get tokens quickly, and you can expect people to pay for that, even if we presume they could run the model locally with no maintenance overhead. There's even cases where using the cloud is cheaper than you can serve locally at all - I saw one guy that measured his added cost of electricity when his computer ran a specific LLM, and it was more expensive to run it locally than to buy tokens from the cheapest inference provider for the same model, presumably because they had better hardware and cheaper electricity.

When you boil it down, situational awareness is little more than our attempt to maintain a mental model of what's happening around us. And models can be wrong.

Do humans have eyes on the sides of their heads? In the back of their heads? Can they watch all directions simultaneously? No.

Can they instantaneously detect changes in speed and direction? No.

Can they react in microseconds? No.

No to all of that. In fact, numerous studies have shown that most people can only deal with about three events occurring at the same time. Throw one more into the mix and they start to lock up.

And "the fact that numerous people surrender all of that" is precisely the problem. People get distracted. They daydream. They drive on autopilot. They become impaired.They lose focus. They get old and slow down. They're not looking in the right direction when some event occurs. And even when they do they often make poor decisions. They overreact and they overcompensate.

And that's when 40,000 people a year die.

Pretty sure most Model 3's and Y's don't have odor detectors, and those are supposedly the vehicles owners would send out to make money why they sleep.

Tesla's numbers already indicate that it's better. (Then again, they're Tesla's numbers.) But my point remains, if they cut down the number of deaths, injuries, and fender-benders it's a net win.

What, if anything I posted was in defense of Elon? All of the above were facts. That's where the HQ is. That's where the engineers are. That's the Texas law.

Musk picked Austin for Tesla’s robotaxi trials and rollouts primarily because the city serves as the home base for Tesla's corporate headquarters, and it offers close proximity to company engineers and executives overseeing the project.

Also, Texas law does not permit individual cities to regulate or ban autonomous vehicles, so it's an AV friendly environment.

Finally, other autonomous vehicle networks, such as Waymo, had already established testing footprints in Austin.

So sorry, but I think your conspiracy theory is a bit short lived.... ;)

Not enough to make up for waking up in the morning to go to work, only to find out some drunk puked in your front seat.

Or the used condoms in the back.

That's one of the reasons I never understood the pushback against EVs in certain circles. Put a solar panel on your roof and you're now independent of the grid and gas stations.

I mean, it's a LOT harder to put an oil well and refinery in your backyard.

They are, for the most part, illegal. That doesn't seem to stop the deaths and injuries, now does it?

40,000 people a year. Year in, and year out.

3,000 people die in a WTC attack and we flip out.

Kill 40,000 men, woman, and kids each and every year... and we accept it as "normal".

I agree wholeheartedly that we should do everything possible to prioritize mass transit, walkability, and biking, and minimize car use.

But.

In the US we've spend the last century designing our cites around urban sprawl. Like it or not, that's where many of our homes are and where people live.

You're simply NOT going to solve sprawl, especially in western cities, with mass transit.

As such, some form of personal transportation must exist. And if so, then it benefits everyone to make it as efficient, and as clean, as possible.

Which doesn't mean that in an emergency that they can't shut down roads and highways to automotive traffic or stop and arrest offenders who're violating the law.

Not to mention that your ability to operate a motor vehicle is largely dependent on the state and deemed a privilege, not a right. (Sovereign citizens to the contrary.) It can be taken away.

Your version of "personal autonomy" is illusionary.

Thank you.

You honestly think an always-on system that's always looking to the front, back, and sides simultaneously and that reacts in milliseconds is going to be less safe than all of the other dumb, drunk, distracted, texting, road raging idiots who currently drive their cars into walls, other cars, pedestrians, bikes, etc.?

In the US we currently kill 40,000 people a year in auto accidents. Injure 2.5 million more. Have 6 million collisions.

That's a pretty low bar to beat. Do HALF as good and you save 20,000 people a year.

People keep writing stories about how self-driving cars might kill them... all while ignoring the fact that our cars (and drivers) are killing about a hundred of us each and every day.

They're reportedly fixing those in future iterations as China balked and passed a law requiring mechanically operateable handles.

`Which still brings an interesting question to light. If self-driving vehicles turn out to be dramatically safer, will insurance companies give reduced rates to them?

Or to flip that around, charge "normal" drivers extravagant rates because they're costing them more money?