What Happens After We Solve The GPU Shortage Problem?
I recently conducted a survey to find out what's going to be the biggest blocker
Welcome to Infinite Curiosity, a newsletter that explores the intersection of Artificial Intelligence and Startups. Tech enthusiasts across 200 countries have been reading what I write. Subscribe to this newsletter for free to receive it in your inbox every week:
I recently conducted a survey on what happens after we solve the GPU shortage problem. Here's the result:
The survey shows that the biggest challenge is the technical knowhow. The cost seems to be the second biggest concern. But the history of compute has shown that the cost always goes down. So this won’t be an issue.
Once we solve the GPU shortage issue, the survey shows that many people don't know how to put all those GPUs to use.
There are two ways to bridge this tech skills gap:
1. Upskill people so that they know how to use GPUs
2. Build a workflow product that makes it easy to use those GPUs
Upskilling people is a big uphill battle. It's usually faster and cheaper to build a workflow product to bridge that gap i.e. make it easy to use and more people will use it.
This is going to be approached from two directions:
Software for GPUs: GPU and compute service providers will continue to build infrastructure software that can sit on their hardware. This will make it easy to use.
Software for people: This is where people build products to make it easy to access all that compute.
What does a good workflow product look like?
It needs to deliver on the following 5 items:
You can access compute with a simple API call
You can access compute across multiple compute providers
You can choose the compute capacity easily and dynamically
You don't have to commit to any kind of contract (short term or long term)
It comes pre-installed with all the necessary packages
For startups, there's a workflow product opportunity here. If you can make it easy for developers / researchers to access AI compute (10x easier than what it is today), then more people would be willing to consume it. It means that there's a much bigger untapped customer base waiting to use your workflow product.
If you're a founder or an investor who has been thinking about this, I'd love to hear from you. I’m at prateek at moxxie dot vc and our fund Moxxie Ventures leads seed rounds.
If you are getting value from this newsletter, consider subscribing for free and sharing it with 1 friend who’s curious about AI: