If we want to trust AI applications, we need to be sure that the data underlying the AI ​​models is reliable. Fujitsu wants to play an important role there, including using blockchain technology.

Fujitsu unveiled its new AI strategy to the world last week. The Japanese technology specialist has been researching artificial intelligence for years and now wants to play a role in the generative AI trend. Since Fujitsu is not alone, the company has to look for special characteristics. According to Vivek Mahajan, Chief Technology Officer, and Ichiro Auyagi, Co-Head of Fujitsu Uvance, trust and reliability will become a priority asset. A second advantage is related to geography: Fujitsu’s Japanese DNA also seems to appeal to customers in distant Europe.

First good data, then good AI

“AI is the final piece of the puzzle to extract value from data,” says Auyagi, putting Fujitsu’s ambitions into context. “The market potential and associated opportunities are also immense, amounting to trillions of dollars. We are now only at the first percent of what is to come. AI will be integrated into everything.”

However, the success of AI depends on trust in it. This trust is inextricably linked to the data behind the AI ​​model. “Trust isn’t actually about the algorithm,” Mahajan notes. “It’s about the credibility of the data processed with AI.” Fujitsu wants to strengthen this trust by combining its own technology with AI. “We can combine AI with blockchain to create trust.”

The power of blockchain is revealed here in new and relevant ways. Don’t think of digital coins, think of professional blockchain technology that tracks bank transactions, for example. There are also things like parts and food that are tracked throughout the supply chain. If AI can make decisions based on such verifiable data, those decisions will immediately be much more reliable.

Looking for access control for LLMs

In addition to the trust issue, there are still some issues that need to be resolved. For example, what do you do with an AI that has access to all company data if it needs to answer a question from someone who has less access? For example, the marketing department should not receive answers from the chatbot within the company based on the most recent quarterly results, while the bot should take these numbers into account when the CEO asks a question.

“The problem has not yet been solved,” says Mahajan. “How do you ensure that people only have access to the data they are allowed to see? There are access control solutions for this in a traditional environment, but how does it work with LLMs? The reality is that there are currently no complete answers.” Mahajan insists that Fujitsu is involved in researching the solution.

Uniquely positioned

In any case, individuality and specialist knowledge are important. Fujitsu also expects to play an important role there. “Certainly multimodal LLMs are a great opportunity for us. AI is being integrated into every aspect of the business, and we are very strong in certain areas. For example, we have a lot of expertise in the manufacturing industry,” says Mahajan. “No company will win AI while the rest will lose. There’s a role for everyone.”

No company will “win” from AI while the rest will lose.

Vivek Mahajan, CTO Fujitsu

Auyagi notes that Fujitsu customers aren’t suddenly knocking on the door and asking what AI can do for them. “You develop a business case; a specific problem that we help solve.” AI can play a role in this.

Own hardware

Fujitsu’s focus on trust and expertise in key vertical markets makes it a relevant party in the AI ​​story, but the company has another strength. Mahajan: “We see many opportunities in Europe. As a Japanese company, we can offer a different perspective there.” He also points out that most of the big AI players today come from the US. Fujitsu is sometimes perceived as a more independent party. “Perhaps geopolitics will work in our favor,” the CTO speculates.

This is what Fujitsu’s new strategy says Decision Intelligence Platform-as-a-Service central. This involves combining building blocks in combination with the trust elements mentioned above to form a basis for AI. But software is not the company’s only strength. With its own Fugaku system, Fujitsu was at the top of the top 500 most powerful supercomputers in the world for a time. The chip behind this computing power was the home-made A64FX. This ARM processor will be derived from Fujitsu Monaka.

“Monaka is a 2nm dual-socket chip with 44 cores,” says Mahajan. “The focus is on electricity consumption. AI and GPTs consume a lot of power. This chip will be very relevant in the ecosystem, not as a competitor to Nvidia or Intel, but for use in sectors such as telecommunications. The new ARM chip is planned for 2026.”

Challenges, development and opportunities

With a software strategy that is linked to a hardware component and both focus on their own strengths, Fujitsu expects to be prepared for the AI ​​boom. Of course there are still challenges. “How to further develop AI is a difficult question,” recognizes Mahajan. “You have to let go of the technology and move forward at the speed it needs to, but for that you need a framework. Governments can play a role here. However, it is not practical to stop development. It moves too fast and knows no national borders.”

Finding the right people will also be a challenge. “Jobs will change, but not necessarily disappear,” Auyagi notes. “Technology is always changing the workplace. People need to retrain, but I think there will be more work. To develop AI we need technical experience, but also business knowledge. When business expertise and technical skills come together, it is a very powerful combination.”

In any case, both men are excited about the future, the role that Fujitsu can play in it and the expected impact of AI on society. “AI is not just a big business opportunity,” concludes Mahajan. “It will always make people happier.”