Muddu Sudhakar: I run a PaaS company. I used to work for VMWare. We had the Hadoop stack and catered to Java. What happened is you have to optimize it to a cloud. You need to take your PaaS and optimize to one infrastructure. If you have a proprietary NLM, it’s not going to be specialized to Azure as well as Azure can.
I, as a vendor, have to make sure that my NLMs will run on more than one cloud provider. Customers may want multiple choices. The burden is on me to make sure that Aisera algorithms can run on AWS, Azure, and Google. I’m optimizing to each of the stacks and not going to 20 of them.
Sramana Mitra: When you look at the top players like the ones that do have heavy-duty cloud infrastructure already that they will augment or are in the process of augmenting, how do you stack Microsoft against Google and AWS? How do you see that battle shaping up?
Muddu Sudhakar: Microsoft has a lead over other teams with ChatGPT and OpenAI. They put the stake in the ground. If you look at what happened with AWS, they came up with the cloud in 2006 to 2008. They are the lead. Azure and Google are catching up.
Microsoft will have the lead for years to come. I won’t be surprised if AWS will launch their version of conversation AI and generative AI optimized on AWS. Right now, Azure and Microsoft have the leads on attracting startups and large companies to run and execute on their stack.
Sramana Mitra: Microsoft is definitely ahead on the generative AI side. Google, however, is ahead on the infrastructure side. Google’s capability and infrastructure for running heavy-duty computing is higher than Microsoft’s. Google and Microsoft are very well-positioned to play this game. They both have their own proprietary generative AI.
AWS and Oracle will have to develop it or acquire. There’s potentially two big acquisitions that may happen in this process. One, AWS acquires an NLM from the market. Oracle acquires an NLM from the market. It’s going to be a four-horse race.
Muddu Sudhakar: All the cloud providers have to have that generative AI framework. What they have to provide is almost similar to what Azure and Google are providing today to make it easy to consume the microservices from them so we can add value and differentiate and build applications. The ecosystem will flourish like how the cloud has done. Companies built on cloud AWS, Azure, and Google. It took almost 20 years. The next 10 to 21 years will probably be the golden age of AI.
Sramana Mitra: The smart strategy for entrepreneurs is to pick one of those infrastructure providers and just build on them instead of trying to do a whole NLM business.
Muddu Sudhakar: You need to figure out what business problem you’re trying to solve. What is the pain point? For that problem, I need to create a full SaaS application. If you’re a technology guy providing a stack and you’re doing some kind of a database, that becomes a margin player.
I see people building databases optimized for ChatGPT, which is a feature that gets absorbed. All the platforms are offering databases as a service. The play has to be SaaS applications or provide the entire PaaS.
This segment is part 3 in the series : Capital Efficient Strategy for Generative AI Startups: Aisera CEO Muddu Sudhakar
1 2 3 4