COVER STORY
Thomas Kurian On Google Cloud’s AI Di erentiators Vs. Rivals AWS, Microsoft The CEO discusses what he says are the advantages Google Cloud has over its two main rivals, the big money-making opportunities for AI MSPs, and how Google is investing heavily to enable GenAI sales opportunities for partners.
What is Google Cloud’s market di erentiation in AI versus AWS and Microsoft? We look at it as five important things. The first one, we built the world’s best AI infrastructure for training and serving models. When I say that: cost, performance, scale, throughput—all metrics. If you’re building an application, you’re going to have half the cost, which is critical for people when they adopt it.To show proof of that, 70 percent of the AI unicorns and 50 percent of all AI startups use our GCP [Google Cloud Platform]. So these are the people who are knowledgeable, and they’re voting with their feet. Secondly, when you look at companies and you look at solution partners like systems integrators, etc., they want a platform. AI is not about just the model, it’s about all the tools you need around the model. What do you mean by that? Imagine you’re building a customer service example and you’re building with a model. How do you make sure the model doesn’t start abusing the customer? How do you make sure when you ask a question about ‘What’s my bill?’ you get an answer [that is] 100 percent accurate 100 percent of the time. You can’t have, ‘Oh, well, sorry.We made a mistake 20 per- cent of the time because our model hallucinates.’ So there’s a collection of services that go around the model. And we give people a platform, based on all the experience we have at Google, that enables them to have all the services around the model. That’s why they’re able to go faster.
third parties. It allows them to choose the model that they want. You know that garbage in is garbage out. So if your data is not clean, no matter what—your model may be an elegant model—but it won’t work. So we’ve got all that as a technology foundation. Four, we have a long history of integrating AI models into our products. Just saying, ‘I’ve got a model’ is not as capable as saying, ‘It’s integrated inside the product.’ Because the experience people have inside the product—How good are your prompts? How many times do I have to ask a question in the prompt to get an answer back?—that’s expertise we’ve built for a very long time. Gmail, just as an example, our productivity tools, we inte- grated AI inside of them 10 years ago. So [we brought] that expertise in terms of how we make models work efficiently inside products to Google Workspace and to Google Cloud Platform using Duet AI. And that enables partners to build solutions much more quickly around it because we’ve done it. Then lastly, unlike the competitors who have large services organizations, we’re not going to customers saying, ‘You have to use our services.’ In fact, we’re telling customers, ‘Start with your systems integrator, the consultant, the technology partner, the AI MSP that you’re comfortable with.’ What we’re going to do is make our services people assist them on the project. That’s a very different approach than saying, ‘Well, I’m not sure if my partner has the expertise so I’m taking the business direct.’ As you see the need for a next-generation AI MSP, what are the money-making opportunities for them? There are three big things that we see. First of all, a lot of his- torical cloud computing technology, infrastructure and analytics were primarily sold in the IT department.With generativeAI, a
What are Google Cloud’s other market di erentiators?
Third, we give them a choice of a variety of different types of models: from Google, from open source and even from
12
DECEMBER
Made with FlippingBook - Online catalogs