How does the head of the world’s No. 1 database company see the current AI trend?

How does the head of the world’s No. 1 database company see the current AI trend?

At Oracle AI World in Las Vegas this week, Larry Ellison, co-founder and CTO of Oracle, was on the keynote stage

The 81-year-old gave a keynote speech sitting on the sofa, perhaps because of his age.

Since it is a place to create business opportunities by boasting Oracle’s technology, I was able to confirm Oracle’s opinions on its products and services and its use of AI and databases.

This paper summarizes his keynote speech, which lasted about an hour and 20 minutes, as follows.

1) “Private Data Is Real Value of AI”…Oracle Strikes Back
2) From medical data to drones…Healthcare All-round Strategy
3) “Preparing for the Food Crisis in 2050” …Investment in Air Pressure Greenhouse and Genetically Modified Crops
4) World’s Largest AI Cluster in Texas…Open AI with 500,000 GPUs

I’ll start today’s article focusing on this.

1. “Private Data Is Real Value of AI”…Oracle Strikes Back

Chairman Larry Ellison argued on the same day that the use of private data held by each company is key for AI models to be truly valuable.

Most AI models currently in service, such as ChatGPT, Antropic, and Grook, have been trained with publicly available data, but the actual business value is maximized when companies utilize their private data.

He emphasized this point as the competitiveness of Oracle.

This is because, globally, most of the high-value data is stored in Oracle databases.

At the same time, Chairman Larry Ellison diagnosed that companies are in a fundamental dilemma.

Individual companies want to keep private data private, but at the same time, they want to infer using powerful AI models.

The solution to Oracle was RAG technology, in which the Oracle AI database vectorizes data, making it accessible to AI models, but maintaining the security of private data.

In addition, he emphasized that using Oracle cloud infrastructure gives you more choice.

You can freely choose and use popular multi-modal models such as Grook, ChatGPT, Lamar, and Gemini.
It is also possible to mix and use several models.

He explained that the scope of data utilization could be expanded. Not only public data, but also Oracle databases, other databases, and even private data on Amazon cloud storage could be vectorized and used for inference.

At the same time, Larry Ellison emphasized that this is not an easy task, and if it was, many companies would have already done it. He reminded Oracle of its differences from major cloud providers such as AWS, Microsoft, and Google.

Oracle directly provides enterprise applications such as medical applications and financial applications.

The goal was not just to create AI technology, but to solve real problems in various industries with it.

Chairman Larry Ellison went on to say that Oracle is already validating this strategy internally.

Oracle vectorized all the customer data and then made it accessible to the multimodal AI model through RAG. And it asked the system a question. I asked them to predict who they are likely to buy another Oracle product in the next six months, and what products each customer will buy.
What’s surprising is that the system didn’t just answer.

AI agents have automated real business processes.

Should I give you an example?

If the prospective buyer is a Swiss bank, the Swiss bank that has already purchased the product is selected as the best reference. The system utilized all the information Oracle knew about its customers. It analyzed the customer’s situation, business content, products they had, and other companies with good relationships.

And I automatically sent you an email containing three optimal references.

All of this is made up of a line of natural language commands.

There have also been major changes in the way they have been developed. Oracle has added a declarative AI generation language to Apex for application generation.

• Declarative programming refers to expressing the logic of computation without focusing on the flow of control.

For your information, most of the new applications that Oracle is currently making are not handwritten.

The created AI agents are connected to the workflow.

In terms of operations, there was a more interesting story.

The quality of AI-generated applications was superior to that of humans.
From a security point of view, there were no loopholes and no mistakes or omissions.

In particular, Oracle Chairman Larry Ellison was confident because all applications were built in a way that did not store data. He explained that even if it stopped working in one computer room, it could be restarted right away in another computer room, no data would disappear, and it would continue to go back without the user’s knowledge.

It is automatically backed up, there is no problem if one location fails, and stability, security, and scale are all built into it.

Finally, Chairman Larry Ellison pointed out the problems of many easy coding tools.

These tools can be used by a team and can be used by a few dozen people, but it slows down when more people use them. Oracle, on the other hand, said it always designs with millions of users in mind, even if there are only five users.

He explained that this is faster to run, uses fewer resources, and doesn’t need to redesign as your business grows.

So how do these applications specifically apply in the real world?
In the next part, we’ll look at this in more detail.

2. From medical data to drones…Healthcare All-round Strategy

Four years ago, Oracle spent about 34 trillion won ($28.3 billion) to acquire Cerner, a U.S. electronic medical record (EHR) company.

At the time