SoftBank's president said he is "honestly exhausted" as cyberattacks on the company reach about 700 million a month, warning that AI is giving attackers new speed and new openings faster than defenders can close them.
The group reported record results for the six months to September on November 9. Revenue rose 11.7% year on year to ¥3.7987 trillion, operating profit 2.1% to ¥642.1 billion and net profit 4.5% to ¥364.5 billion.
SoftBank raised its forecast for the year to March 2027 to ¥7.9 trillion in revenue, ¥1.16 trillion in operating profit and ¥610 billion in net profit.
Cloud and AI carry the half
The enterprise business, which covers corporate network services, cloud, security and AI services, led the gains. Revenue there rose 11.8% to ¥538.9 billion and operating profit 19.9% to ¥124.8 billion.
Revenue in the cloud and AI segment, which includes AI computing infrastructure, grew 36.7% year on year. An AI data center in Tomakomai, Hokkaido, is on track to open during the fiscal year ending March 2027.
Consumer services such as mobile and power brought in ¥1.6025 trillion, up 8.6%. Media and e-commerce, which includes advertising on Yahoo Japan and LINE, rose 8.7% to ¥894.7 billion, and finance services such as cashless payments rose 27.5% to ¥241.9 billion.
An AI arms race
More than 7,000 companies have signed up for Patching as a Service, which helps clients fix vulnerabilities in their IT systems. President Junichi Miyakawa said the work has to run a fast, continuous cycle of three steps.
- Diagnose vulnerabilities
- Apply the fix
- Confirm the system is running normally
The service uses AI technology from OpenAI. Applying patches is automated, but checking that systems run normally is still done by hand, and SoftBank is developing toward full automation.
"Attacks using high-performance AI keep increasing almost every week," Miyakawa said. "The ability to create entry points and the speed of an attack grow stronger by the day."
Each new AI model released brings previously unknown vulnerabilities to light, he said, putting systems that were already dealt with at risk of being hit again.
"As long as attackers' AI keeps evolving, we on the defense side have to keep evolving with AI too," he said.