Motoaki Saito, president of Tokyo supercomputing start-up Pezy Computing, is a man on a mission.

"There is no point coming in second," the 49-year-old whiz said in a magazine interview last year.

"The first country to run next-generation supercomputers will gain unparalleled advantage and reap the benefits," he added, lamenting that Japan could potentially end up being five years behind China on this front if not enough attention is devoted to developing speedier and stronger machines.

But perhaps Saito, who is renowned for his work in downsizing energy-saving supercomputers, has been too ambitious.

Last month, he was charged with defrauding the government of 431 million yen (S$5.2 million) in grants, while a probe is ongoing into several other companies that are linked to him, which are said to have benefited from more than 10 billion yen in subsidies.

On top of that, Saito's Pezy Computing is now being accused of evading corporate taxes, media reports said this month.

His arrest sent shockwaves across the industry, coming weeks after the global TOP500 league table, which measures theoretical supercomputer processing speeds, ranked his Gyoukou as the fourth fastest in the world.

It had shot up 65 places, from 69th, in the previous chart in June last year. The Gyoukou can, theoretically, run at about 19,140 trillion calculations per second or 19.14 petaflops. He wanted his machine to surpass China's Sunway TaihuLight, the world's fastest at 93 petaflops, within a year.

Saito's ambition symbolises Japan's intense rivalry with China in the race to build faster computing machines with stronger brute force, with Japan having ceded much ground to the likes of Silicon Valley and China's multiple tech hubs from Beijing to Shenzhen, over the past decade.

But the official view has been that speed, the Holy Grail of supercomputing, is not the only thing that matters, by any stretch.

A spokesman for the Ministry of Education, Culture, Sports, Science and Technology (Mext), which is behind Japan's supercomputer efforts, said: "I will withhold from commenting about China's supercomputers, the TOP500 measures but one aspect of the performance of supercomputers, with speed as the basis.

"We see as extremely important factors such as power consumption performance, computing capacity, ease of use for the end user and the possibilities for innovative research."

Fujitsu, an industry leader in supercomputer development, is now involved in building a new machine known as the AI Bridging Cloud Infrastructure (ABCI), which will be ready this year and used for artificial intelligence research.

A Fujitsu spokesman said the ABCI would rank third on the most recent TOP500 rankings, and have the honour of being the fastest machine in Japan.

Yet the Ministry of Economy, Trade and Industry (Meti), as the key driver behind the ABCI project, rejected any attempt to refer to the ABCI as a "supercomputer".

A spokesman stressed to The Sunday Times: "It is very important that you understand the ABCI is not a supercomputer with the aim of having the world's fastest specifications. It is a high-performance computer solely dedicated to AI research and development."

The machine cost 19.5 billion yen and will be tapped for research in areas such as productivity and healthcare, said Mr Fumito Sasaki of the ministry's research and development division.

A spokesman for Mext, meanwhile, said that Japan hopes to harness the huge potential of supercomputers to map the impact of Japan's societal problems, as well as to seek solutions through means such as deep learning and conducting AI simulations.

Such issues, the spokesman said, could include its ageing demographic, disaster risks, energy use and low productivity.

With such urgent aims in mind, the ministry has spent big on supercomputer research and development. Mext invested 111.1 billion yen on developing the K computer between 2006 and 2012.

Another 111 billion yen was devoted in 2014 to developing what is known as the Post-K supercomputer that Mext wants to succeed the K by 2022.

The race has also been heating up on the quantum computing front.

Quantum computers differ from conventional supercomputers in that they rely on theoretical particle physics and run on subatomic particles such as electrons in sub-zero temperatures.

Telecoms firm NTT in November unveiled Japan's first quantum computer prototype, with a potential processing speed of 100 supercomputers but using only 1 kilowatt of power - or about what is required by a large microwave oven.

This is a major breakthrough for Japan in the area, as quantum computers have proven to destabilise easily and are error-prone, thereby limiting their functions.

Meanwhile, other Japanese tech leaders, such as NEC and Fujitsu, are also working to develop their own quantum computers.

Mext said Japan will, moving forward, go big on quantum computing research as it "could lead to a society and productivity revolution".

Still, there are concerns that Japan's efforts will hit a snag with Saito's arrest, given his pivotal role in the computing industry.

The Sankei Shimbun quoted an official as saying: "It would be a shame for Japanese supercomputer development to stall because of this recent incident."