GPUs & Compute

AI and decentralization have been converging over the last few weeks.

Meaning- take, for example, the CEO of Stability AI resigning to work on decentralizing AI .

And now, Bitcoin miners are looking at GPUs and AI.

1/ AI intersection with Bitcoin Mining

AI is the next target of many data centers since the bitcoin halving.

For those who don’t know, the reward for mining bitcoin has decreased from 6.25 BTC to 3.125 BTC.

To adapt to this, many mining companies such as Hive, BitDigital, etc are planning to operate AI. And to remain competitive, many miners are now setting up in areas with lower electricity costs.

The intersection of motivations between bitcoin miners and GPU centers is fascinating.

2/ Aethir and IO Net to combine 640,000 GPUs to decentralize AI Infra

Speaking of decentralization, companies with GPUs are finding ways to achieve the goal of decentralization goals. 

Basically, the demand for GPU compute is growing fast, and is projected to quadruple by 2030- this collaboration aims to provide a highly scalable solution for this rising GPU compute all over the world.

3/ SF Compute to the moon?

Speaking of the rising demand for compute, a startup is attempting to democratize this by renting out GPUs to startups.

Yep- the SF compute company is allowing startups and researchers to rent out GPUs by the hour- a compute marketplace of sorts.

It’s a great model and really democratizes one of AI’s scarcest resources.

4/ Elon & NVIDIA Chips

Elon disclosed his companies possess 35,000 H100 chips- and with them, he is using the compute for the Optimus Bot, Robotaxi, and a supercomputer named Dojo.

Elon’s always cooking with AI- and the rise in GPU demand illustrates this.

Compute is a super scarce resource- demand is growing exponentially. And so managing this resource will be key in the world of AI for the foreseeable future.