The most compelling argument for SpaceX and Tesla merger is that they share the same core focus: SI and energy. Cars, robots, satellites, are products that sell to fund the advance of SI and product of energy at scale. The ultimate goal: Kardashev Type 2 + Mars. https://t.co/QFhL9cbU8K

Meta made a “deal” with Zuckerberg to issue non-voting shares that he can use to pay the 5% wealth tax. Thus Zuckerberg spread his bets across the political divide. So machiavellian. The fake threat “I will leave California if the wealth tax is enacted” gives away the cunning plan.

Meta increasing compute scale to a cluster of 1m GPUs by connecting buildings with a different generations of GPUs as if it was one big unit. Keep an eye on the proliferation of intelligent methods of using more efficiently existing infrastructure instead of new builds.

Jensen Huang looks more and more like the CEO of Nvidia VC Enterprises.

They laughed at Meta spending lavishly on top dogs.

New architecture. Intel out. Nvidia in. Still Windows 11!

Judging by the time it seems TSMC as the undisputed leader of fabs overplayed their hand.

Marvell sees extraordinary growth in demand for data and compute in the years ahead. And this doesn't include SI space cloud which is orders of magnitude higher. You cannot not be an optimist. https://t.co/IRIuORJuWb

At this scale, it is clear that these things don't just happen. There is a plan here and there is a huge pot of money funding this. One then has to ask: what is the expected return from this investment?

there is no slowing down to the Anthropic and OpenAI safety show. the "omg we can't control our agents" genie will not get back into the lamp.