nharada

You're right, I can't find any primary sources for this number. Yahoo[1] reports this number and attributes it to Bloomberg NEF but I can't find an actual article from Bloomberg with this number, or any actual target number in it.

[1] https://www.yahoo.com/news/china-10-sodium-ion-battery-16505...

You're right, I can't find any primary sources for this number. Yahoo[1] reports this number and attributes it to Bloomberg NEF but I can't find an actual article from Bloomberg with this number, or any actual target number in it.

[1] https://www.yahoo.com/news/china-10-sodium-ion-battery-16505...

Oh yeah I get what you’re saying but is that a thing? Like CATL has had lines before that aren’t making a profit?

What makes you say that? I don’t know this space very well

CATL is claiming mass production of their sodium-ion batteries starts in December, with a target price of $10/kWh. If that ends up even partway true it'll completely change the economics of power storage.

> Price is the primary decider

That's not even true for normal bicycles. Serious cyclists can spend massive amounts of cash on their rides.

Aside from that, if you're using this as a vehicle you want more than just the cheapest thing out there. Reliability and serviceability are important when not having the bike means you can't get to work or your kid's school.

How come I never see any concrete proposals for how to equitably distribute the wealth of AI? It's always either "stop AI immediately for the sake of our labor" or "don't worry sometime in the future everyone will live in utopia probably".

Here's a starter example: any company whose main business is training AI models needs must give up 10% of their company to a fund whose charter is long-term establishing basic care (food, water, electricity, whatever) for citizens.

I'm sure people will come at me with "well this will incentivize X instead!" in which case I'd like to hear if there are better thought out proposals.

If this was effective on humans I think most people would accept having treatments 2x a year

Not to mention that insurance is crazy with such high vehicle values. Easily another $200-300 a month, and that's if you're a good driver over 30 with no accidents.

Four of the worst five wildfires in the US (in terms of damage) took place in the last decade in California. The fire in LA this year is estimated to have cost more than the other four combined.

I assume there's at least one generation of re-sale value. There's a lot of smaller companies that could use GPU capacity and would if it was cheaper. I wonder how the math works out though -- are power requirements such that the actual costs aren't worth it? How long before the chips just get thrown away?

I agree with this 100%, it's not just gambling it's that the idea there's a steady path you can follow to success has basically disappeared. In a survey[1] asking how much money was required for financial success, Zoomers averaged $10MM, almost double millennials and 10x boomers.

A lot of people online took the opportunity to criticize zoomers as out of touch and financially illiterate, but I think most people under 30 have looked at the trends over their lifetime and determined that their lives are going to have so much volatility they need a massive amount of money to weather the storm.

[1] https://fortune.com/2025/01/20/gen-z-9-5-million-financially...

I find this a cynical and possibly incorrect take. I'll admit, if your goal is to become a billionaire and command vast sums of cash from your many vacation homes I'd agree with you, but financial freedom and an interesting life is absolutely within reach without having to take a career path that doesn't bring you fulfillment. Those CTOs that get pushed out once the company reaches their series C almost certainly achieve financial freedom.

This may be your option for maximum control and leverage, but do remember that your day to day will be different, especially as the company grows. As someone who enjoys technical/product work and doesn't particularly enjoy fundraising, sales, or marketing, being the CEO is actually quite a trade-off for my personal fulfillment.

I suspect ads would be an attempt to monetize the free users not people paying $200/mo for Pro. Though who knows...

Gemini has always been the leader in multimodal work like images and video, I expect this won't be any different but am interested to see how it is

So how do they get this data to the GPUs now...? Just run it over the public internet to the datacenter?

I do really like the Unix approach Claude Code takes, because it makes it really easy to create other Unix-like tools and have Claude use them with basically no integration overhead. Just give it the man page for your tool and it'll use it adeptly with no MCP or custom tool definition nonsense. I built a tool that lets Claude use the browser and Claude never has an issue using it.

it might be functionally correct but if you wrote it yourself it could be orders of magnitude faster

I'm stealing "Non-AI Summary"

Roku tho

Musk has shown repeatedly with *this specific product* that he's willing to compromise quality for his ego. Trust is earned.

> While conventional lithium-ion batteries are limited to a maximum range of 600 km, the new battery can achieve a range of 800 km on a single charge

I know it's a mass-market site but this level of reporting is just insulting. Here are the actual numbers from the paper:

> We demonstrate lithium metal batteries achieving a 5–70% state of charge (SoC) within 12 min over 350 repeated cycles at a 4C (8.4 mA cm−2) charging rate, as well as high-energy designs delivering projected energy densities of 386 Wh kg−1 reaching a 10–80% SoC within 17 min over 180 cycles.

My biggest annoyance is that people aren't transparent about when they use AI, and thus you are forced to review everything through the lens that it may be human created and thus deserving of your attention and benefit of the doubt.

When an AI generates some nonsense I have zero problem changing or deleting it, but if it's human-written I have to be aware that I may be missing context/understanding and also cognizant of the author's feelings if I just re-write the entire thing without their input.

It's a huge amount of work offloaded on me, the reviewer.

One thing I really hate is that some companies with poorly design customer service flows actually REQUIRE you to read a code they text you over the phone to a rep.

At least now more companies include a "never read this over the phone" note in their authentication texts.

This already exists for some things, i.e. arcade.ai

Congrats to the team! It's no small feat to launch to the public in this space, and from the amount of testing I've seen Zoox doing it certainly seems like they've put in the work. Best of luck!

Yeah speed is definitely a weakness right now. One option is to switch the models around and try a faster one, that will probably give you some speed-up at the potential expense of accuracy. You can do that in the .skipperrc file. There's some gains to be made for prompting most likely, as well as in the Omniparser model.

A more future thing I'd like to do is have a model specifically for the action selection task which could be a much smaller network. That would really improve the latency.