“I believe so. And I believe, for example -- I mean I think you'd find different points of view from different engineers as we move Fusion to Autonomous Database. I think the cost savings -- our cost -- our cloud cost savings will be around 50%.”

In context

“l of automation, that level of autonomy. It allows us to get much better margins in our database business, in our SaaS business and the rest of our cloud business. Our clouds are more automated, so we have very low labor costs. Our networks are much more efficient, the RDMA networks run so much faster. If you run twice as fast, our costs go down by half. And our networks are much faster than the other clouds. So we think our potential as we scale, our potential to deliver much better margins they're currently delivering are very real. John DiFucci: So is it safe for me to assume that then the upside you keep putting up and you hopefully will continue to do, will add profit and that profit will actually increase as a percentage of the margins will also increase over time? Lawrence Ellison: I believe so. And I believe, for example -- I mean I think you'd find different points of view from different engineers as we move Fusion to Autonomous Database. I think the cost savings -- our cost -- our cloud cost savings will be around 50%. That's what I believe. Now it might be 40%, it might be 35%, but there will be huge cost savings from where we are now, and that's across the entire base of Fusion customers. So that's just one example of how we're using faster networks, faster databases, more automation to make our products more secure. And I keep emphasizing the security is really the primary goal. But as a second order effect, we also end up spending a lot less money to run those data centers. John DiFucci: Great. Thank you, Larry. Thank you, Safra. Operator: Your next question comes from the line of Mark Murphy with JPMorgan. Your line is open. Mark Murphy: Thank you very much and congrats on the great performance. Larry, how do you envision the market transitioning from the AI training phase to the AI inferencing pha”