The have a whole thing about doing their own pronunciations because using the original one from the language they are borrowing from is apparently either pretentious or racist depending on how superior they feel to the region the word originates from.

Volkswagen of all people.

After this fuckup: https://fahrplan.events.ccc.de/congress/2024/fahrplan/talk/Q8ZAV9/

They should be the last ones to deny anyone else using privacy preserving tools.

(Hint the player has multiple audio tracks, there is a dub in English)

Then I would recommend Haven

Are you genuinely having more fun now than with good games from 10years ago? Even 15years ago??

On the contrary, I’m still playing those games sometimes. At the moment it’s Need for Speed: Most Wanted from 2005.

And recently indie games are growing in popularity, those are often quite simple visually, or go for a retro style. Megabonk for example, or Mewgenics or Slay the Spire 2.

This is brilliant, assuming a deity who doesn’t mind outsourcing of prayers, and grants buffs!

For Masayoshi Son, the Boston Dynamics exit looks small beside SoftBank’s current AI infrastructure campaign. The Wall Street Journal reported in April that SoftBank is forming Roze AI, a new venture meant to use artificial intelligence and robotics to build physical infrastructure, including data centers. Tom’s Hardware, citing the Financial Times, reported that Son is aiming for a $100 billion valuation for Roze and a public listing as soon as this year. That puts the $325 million Boston Dynamics proceeds in perspective. SoftBank is not walking away from robotics as an idea. It is moving toward robots as part of the AI buildout, tied to data centers, energy, land and construction.

To me that sounds like it might really be their plan. But it also seems like a shit plan.

They are already on the hook for such an insane amount of datacenter debt, now they sink even more debt into robots under the theory those might help with the first batch of debt? The timeline will never work out. Until autonomous robots are helpful on chaotic construction sites, maybe in 5 years at the soonest, the LLM craze will have gone bust.

Yeah sure I also think that he is a trillionare at the moment. I was just trying to show that it’s absurd to say he has the money even if the stock price falls because he’s already liquidated it.

By that logic he’s not a trillionaire at all. They only sold 4% of the company for some 68 billion USD. A trillion dollars was never liquidated.

Edit: 4.3% and 75 billion are the correct numbers.

Ed’s article where he released those numbers: https://www.wheresyoured.at/exclusive-openai-financials/

Which column is infrastructure under?

I don’t think they have any of that. They rent computation capacity, which should partially be under cost of revenue and partially under R&D.

a picture of a screwdriver on one and a picture of a wrench on the other

Something that screws and something that sounds like wench. I would think huh fun little riddle and then go in. The idea that it was supposed to be ungendered wouldn’t even occur.

Is there a term for “lying by imprecise wording likely to be interpreted in a more serious way that what is actually the supportable message”? Because this headline is such a prime example.

When you read this it sounds like Grok fired missiles in some way. But the actually supportable message is that it was used in target selection, and its speed enabled the missile attack to reach such a scale in a short time frame.

Edit: Not that I think that’s good or anything, don’t get me wrong. I’m just railing against the shoddy journalism, despite the horrifying content.

Different idea, how about we get the cars out of the way of the children again, and also shun LLMs?

I guess the earliest would be in July, because they also stated in the same document I linked:

The Company currently expects the Merger to close during the third quarter of 2026.

It “was” not anything, as it is still upcoming.

They will take a volume weighted average closing price of the seven previous days:

At the effective time (the “Effective Time”) of the Merger, each share of Cursor’s common stock and each share of Cursor’s preferred stock outstanding immediately prior to the Effective Time of the Merger will be automatically converted into the right to receive shares of the Company’s Class A common stock based on an implied equity value of Cursor of $60.0 billion and the price of the Company’s Class A common stock equal to the volume-weighted average closing price thereof over the seven consecutive trading days immediately preceding the closing of the Merger (the “Merger Consideration”).

They are also claiming to be training their own new model on SpaceX’s infrastructure. I don’t particularly trust them, but I’ll give the citation anyway just for additional info.

Together with SpaceXAI, we’re training a significantly larger model from scratch, using 10x more total compute. With Colossus 2’s million H100-equivalents and our combined data and training techniques, we expect this to be a major leap in model capability.

They entered into an agreement to buy Cursor, with an expected time for the merger in Q3. It’s an all stock transaction, the number of shares that will depend on the SpaceX share price, to make up the 60 billion USD. Also they will pay in Class A shares (so the shitty ones with nearly no voting power)

For real, or just for a little pump and dump action by some bad actor again?

It’s slightly less stupid than data centers in space I suppose. But I still find it pretty weird.

You will have to be tethered to land anyway. Properly high bandwidth networking as you would usually see to connect a data center (on the order of a dozen terabits per second) only exists through fiber optic cables. I’m sure of this point, because optical networking is my day-job, though we only run 400 Gbit/s links on the fastest edges since we’re a small national network.

As for power, well there are 80 MW ship engines (e.g. Wärtsilä-Sulzer RT-flex96C, which has even been built in Korea under license before), so it’s not impossible I suppose. But if you are tethered, then the country you’re tethered to will probably forbid you from burning bunker fuel for 80 MW on its coast. At which point you’d be reduced to running clean diesel or something. That I expect would make the power more costly than just tethering to an electric grid

So now we have a big barge rather than a ship. What do you really save then? I guess the price of the land? And you gain access to copious amounts of saltwater, so you can do closed loop cooling with freshwater, and do the secondary heat exchange to the ocean. But you could do that by building on the coast too. Okay I guess you might gain tsunami security over a coastal building.

If this is a real proposal why don’t they tell us the material advantages they expect, rather than making us guess?

der8auer already explained that it won’t work. So this is no big surprise to me, but nice to see him proven correct.

Links: German: first video, follow-up video, English: first video, follow-up video