Sytam

Advisor | WEB3 research | Sharing tips | Growth Producer

[Sytam @by_Sytam]

@Ravikar_Ai access to the tools is only half the equation. the real challenge is making sure the value they create isn't concentrated in the hands of a few.

[Sytam @by_Sytam]

Elon Musk literally explained why he thinks robots will do what charities could not.

> He said it a year ago in Washington, on stage next to Jensen Huang at the U.S.-Saudi Investment Forum.

- The host asked how he keeps disrupting industries. Musk pushed back on the word: "it's mostly not disruption, it's creation." (01:25)

- There were no reusable rockets. You could not buy an electric car, he says. And right now there are no useful humanoid robots either. "There are sort of gimmicks." (02:36)

- Tesla will make the first real one, in his telling. Then the big claim: humanoid robots will be "the biggest product ever," bigger than cell phones, because everyone will want one. Maybe more than one. (03:45)

- Then he turned to the people who have promised to end poverty for decades. Lots of NGOs trying, he says, "really not succeeding." (05:50)

- His alternative: AI and humanoid robots "will actually eliminate poverty." Tesla will not be the only maker, he adds.

"There is only basically one way to make everyone wealthy, and that is AI and robotics."

> That same day, he announced a 500 MW xAI data center in Saudi Arabia.

A year later, Tesla has started early Optimus production in Fremont. Musk says 20,000 robots a week is still at least a year away.

[Sytam @by_Sytam]

@ExploringSpace2 maybe the goal shouldn't be a world without work, but a world where people have more freedom to choose what makes their effort meaningful.

[Sytam @by_Sytam]

@cassian0x if it actually works out, yeah. labor that doesn't sleep and gets cheaper every year changes the math for pretty much every industry

[Sytam @by_Sytam]

@dexmarlow agreed, big claims are cheap, weekly output is what shows if it's real. if they actually get to 20k a week the cell phone comparison stops sounding that crazy

[Sytam @by_Sytam]

@mak1tra Most people would have sold at -30%, let alone -83%

[Sytam @by_Sytam]

@sterlingmkts Skill gets you through the door, but surviving those first two years with $70k revenue against $160k overhead is pure stubbornness

[Sytam @by_Sytam]

@awi_digest Yes, but that’s a matter of chance; elections are currently underway in America, and the market is reacting to that.

[Sytam @by_Sytam]

Chamath Palihapitiya literally explained why the market now prices CoreWeave's debt like a coin flip.

" The AI bubble talk is the cover story. "

> He walked through it on All-In, with Ben Horowitz filling in.

- Founder Michael Intrator explained the box on the show in March. A client contract goes in. GPUs from Jensen go in. The data center deal goes in. (02:18)

> Chamath's point: the box still runs on borrowed money.

- CoreWeave has about $35 billion of debt, he says, and needs roughly another $35 billion to fund every contract it signed. It pays about 8.5% on average. One GPU loan spiked to 15% that week. (07:23)

- The contracts are fixed price. Rates go up, revenue does not. And 75% of revenue comes from three companies. (09:39)

- Default insurance on that debt hit about 953 basis points. In his math, that is a 16% chance of default in any year and about 55% over five. "That's not what I'm saying," he adds. (11:24)

His team's estimate: CoreWeave sits about $5 billion from break-even.

> He does not call it a real risk yet. $78-80 billion of assets against $35 billion of debt. But he calls it "a confidence

[Sytam @by_Sytam]

@theSethian what do u mean?

[Sytam @by_Sytam]

@skunzu same lol. napkin version: a radiator at room temp sheds roughly 1 kW per m2 at best, so a terawatt means something like a thousand km2 of radiators. he didn't really show that part

[Sytam @by_Sytam]

Elon Musk literally explained why the AI race will not be won on Earth.

" Forbes asked him at a dinner in May when the first data center goes to space. "

> His math starts with how small we are. All of human civilization uses much less than a trillionth of the sun's energy, he says. "Everything is irrelevant compared to the sun." (00:25)

- So the fights over this or that power plant miss the point, in his telling.

- Then the plan. SpaceX has about 10,000 satellites in orbit. With Starship, he says, it will launch over 10,000 communication satellites a year. (02:05)

- His rough estimate: a terawatt of solar-powered AI satellites launched every year. About 10 million tons to orbit, at 100 kW per ton. (03:10)

- The average US power use is half a terawatt. So every year, he says, SpaceX could put up twice the power America consumes, just for AI. (04:13)

"We're not breaking any physics here."

- Step two is the moon. A base, a mass driver, solar panels and radiators built there. He puts that at a petawatt a year, and says Earth's economy would look inconsequential next to it.

In October, Google's first Project Sun catcher satellite reached orbit with four of its AI chips on board. It flew on a SpaceX rocket.

[Sytam @by_Sytam]

@OxPips yeah that's the part he kinda glossed over. in vacuum you can only radiate heat away, so the radiators get huge. makes sense why he mentions building them on the moon

[Sytam @by_Sytam]

@diamai_ lol imagine opening a ticket and waiting for the next starship window to get one gpu swapped

[Sytam @by_Sytam]

@Z0llan Curta in pure HTML? Wild.

[Sytam @by_Sytam]

@VVMotyka Demos to daily workflows, finally

[Sytam @by_Sytam]

@0xLonkir https://t.co/LTzeM2X3qK

[Sytam @by_Sytam]

@Stiingy7 u can just turn on video during trip to work, or from

[Sytam @by_Sytam]

@0xLonkir i can send you link of interview, would you like?

[Sytam @by_Sytam]

@explosss1ve Foreman with 103 tests against lying bots is the real gem here