As I mentioned yesterdayJensen Huang, co-founder and CEO of Nvidia, opened his company’s GTC conference with several product and service announcements, including introducing the two cloud computing services that the company will run.
At a press conference on Wednesday, Huang told ZDNET that these two services will be “extremely long-term SaaS platforms for our company.”
single service , Big language model cloud servicesLet’s take an AI deep learning software developer like Nvidia’s GPT-3 or Megatron-Turing 530B, and tune it to specific applications, to make it specific to a task while reducing the effort the customer has to do.
second service, Omniverse Cloud Servicesis a rendering of Nvidia’s Infrastructure-as-a-Service that will allow multiple parties to collaborate on 3D models and behaviors.
I asked ZDNET Huang, what is the size of SaaS [software-as-a-service] Be working for Nvidia over many years?
It was hard to tell, Huang said, but the large language model service has such a wide application, and it will be one of the biggest opportunities of all programs.
Here is Huang’s full response:
Well, it’s hard to say. This is really the answer. It depends on the software we offer as a service. Perhaps another way to take it is just a couple at a time. We’ve announced this GTC, new chipsets, new SDKs, and new cloud services. This is what you are asking about. She highlighted two of them [cloud services]. One of them is the large language models. And if you haven’t had a chance to look at the effectiveness of large language models and their implications for AI, please really do. It’s really important stuff. Large language models are difficult to train, and the applications of large language models are quite diverse. He has been trained on a great deal of human knowledge. And so it has the ability to recognize patterns, but it also has an encoded amount, a large amount of encoded human knowledge, so that, if you will, it has a kind of human memory, if you will. In a way, much of our knowledge and skills have been encoded. And so, if you want to adapt it to something it hasn’t been trained to do – eg, it was never trained to answer questions or was never trained to summarize a story or make breaking news, paraphrase, was never trained to do These things – with a few extra shots of learning, you can learn these skills. This basic idea of fine-tuning, adapting to new skills, zero-shooting, or a little bit of learning has major implications in a large number of fields, which is why you see such a large amount of funding in digital biology. Because large linguistic models have learned to structure the language of proteins and the language of chemistry. And so, we developed this model. And how big is this opportunity? My sense is that every company in every country that speaks every language has probably dozens of different skills that their company can adapt our big language model to start performing. I’m not quite sure how big this opportunity is, but it’s probably one of the biggest software opportunities ever. The reason for this is that intelligence automation is one of the biggest opportunities ever.
The other opportunity we talked about was Omniverse Cloud. And remember what an omniverse is. Omniverse has many characteristics. The first characteristic is that it absorbs, it can store, it can synthesize physical information, three-dimensional information, across multiple layers or so-called schemas. It can describe geometric shapes, textures, materials, and properties such as mass, weight, and the like, contact. Who is the supplier? What is the cost? What about? What is the supply chain? I would be surprised if – behaviors, motor behaviors. They could be AI behaviors. And so, the first thing Omniverse does, is it stores data. The second thing it does is that it connects multiple proxies. And customers can be people, they can be robots, and they can be autonomous systems. And the third thing it does, is it gives you a viewport to this new world, another way to say, the simulation engine. Thus, the Omniverse is basically three things. It is a new type of storage platform, it is a new type of communication platform. It is a new type of computing platform. You can write a request above Omniverse. You can connect other applications through Omniverse. Like, for example, we showed many examples of Adobe’s connection with Autodesk applications connected, you know, with various applications. And so, we connect things, and they can be connecting people. It could be connecting worlds, it could be connecting robots, or connected agents. Thus, the best way to think about what we did with Nucleus [Nucleus Cloud, a component of Omniverse Cloud, is a facility for developers to work on 3-D models using the Universal Scene Description specification], think of it as the easiest way to monetize it, maybe it’s like a database. Hence, it is a modern database in the cloud. Except for this 3D database, this database connects multiple people.
And so, these were two SaaS applications that we rolled out. One is called the Great Language Model. The other one is basically Omniverse or a database engine, if you will, we’ll put it in the cloud. So, I think those two announcements – I’m really glad you asked – I’m going to get a lot of opportunities to talk about it over and over again, I’m going to talk about it over and over again, but these are going to be two SaaS platforms that are long-term SaaS platforms for our company, we’re going to make it work in multiple clouds and so on .