🔧 Ask someone outside Japan what "Japanese manufacturing" looks like and you will usually get some version of the same picture: zero defects, kaizen, a shop floor where mistakes are unthinkable. Which makes it a little jarring that when Toyota's chief technology officer was asked what actually drives development speed, his answer was failure. Not tolerating failure. Producing it, in bulk.

Toyota's answer was "the number of failures"

Hiroki Nakajima, vice president and CTO of Toyota Motor, gave a solo interview to Nikkei xTECH in August 2026. The headline distills his answer to a phrase: the engine of development speed is an overwhelming number of failures.

The reasoning isn't mysterious. If getting to a working design means trying, breaking, and fixing, then the team that runs the most cycles arrives first. But it is an expensive thing to say out loud, because attempts cost money and calendar time, and someone has to sign off on both.

Toyota is signing off. Its plan as of August 2026 puts R&D for the business year ending March 2027 at 1.6 trillion yen, about $10 billion, up 5% on the previous year and the highest figure in the five-year series the company publishes. Nikkei xTECH puts the combined total across Japan's seven carmakers at 4.022 trillion yen, roughly $25 billion, which it reports as a record for the group.

There is a reason for the urgency, and it now parks in Japanese driveways. Chinese automakers are commonly said to develop a new model in two to three years against four to five in Japan, the US and Europe. In July 2026, BYD put a kei-class electric car called the RACCO on sale in Japan, priced from 2,145,000 yen (about $13,400) and engineered to Japan's kei regulations. Foreign-brand cars have squeezed into the kei size limits before, but this is reported as the first EV designed for the standard from the ground up by an overseas maker. The Japanese motoring site carview! reported that the project took about two and a half years from concept.

Toyota has also set up a squeeze for itself. The company changed presidents on April 1, 2026, promoting former CFO Kenta Kon, whose stated priority is lowering the break-even volume so the business can hold up when conditions turn. Spend less on running the company; spend more on getting things wrong. Those are the same sentence read from opposite ends.

A 35 billion yen building for getting things wrong

Encouraging failure in a speech is free. Murata Manufacturing did something harder: it paid for the room.

In March 2026, Fukui Murata Manufacturing opened the Ceramic Capacitor R&D Center, known internally as C4-Lab., in Echizen City, Fukui Prefecture. It is, the company says, the first site Murata has dedicated entirely to research on multilayer ceramic capacitors, the tiny passive components that sit by the thousands inside phones, cars and AI servers. Construction began in November 2023 and finished in February 2026. Land and building together cost about 35 billion yen, roughly $220 million. The structure is five floors that stand as tall as a typical nine-story block, with about 530 people working there as of June 2026 and room planned for 800.

Until now, Murata's MLCC research borrowed time on mass-production lines. Sales have been strong, the main plants are busy, and busy plants do not hand over a line so an engineer can try something that probably won't work. So the experiments got squeezed out.

C4-Lab. produces nothing. It has a full prototype line covering every step of MLCC manufacturing, evaluation equipment, and rooms set aside for joint work with partner companies, and none of it is attached to a shipping schedule. The executive who oversees production-technology development for the ceramic capacitor business told reporters at a June 2026 press tour that the site was designed to absorb risk-taking rather than punish it, and that with new ideas, in his words, how much you can fail is what decides it.

That is a strange sentence coming from a company whose entire commercial reputation rests on parts that never fail. But failure means different things on the factory floor and in the lab. A part that dies on the test bench is a data point you paid for.

The mid-sized version: procedure instead of budget

Toyota can buy attempts with money. Murata can buy a building. Most manufacturers can do neither, and a company like Metal Technology Co. has to find a third way.

MTC began in 1960, when a metallurgy research group at Japan's national RIKEN institute set up a furnace in Tokyo. It does contract hot isostatic pressing: squeezing metal parts with high-temperature gas at extreme pressure to close internal voids. It is unglamorous and it is everywhere, in jet engine castings, fusion components, 3D-printed parts. The company runs more than 20 HIP units, including one it describes as among the largest in the world, at its Himeji plant.

MTC publishes a roundtable on its own website under the heading "a culture of learning from challenge and failure" (page last updated 5 August 2026). Company-published material deserves a raised eyebrow.

The mechanism is paperwork. MTC runs an internal research contest; the top three proposals get roughly two years of support, and the researcher's own department redistributes their day job so the work can happen. Progress is checked at set milestones through what the company calls gate reviews.

The example they give is a young engineer at the Himeji plant who proposed re-testing something everyone in the field already believed, about how air is removed from a capsule before pressing. His own review committee did not like it. He recalls being asked whether the study had any point, and whether he was simply restating common knowledge. It took half a year of rewriting before the proposal was approved. The finished work was presented at the 14th international HIP conference in Aachen, Germany in 2025 and picked up by the journal Powder Metallurgy. He says the drive came from a conference presentation he had botched as a student.

Even at the company that puts failure in its values statement, the gate is real, and it swings shut hard. The Himeji plant manager says the biggest problem is being too afraid of failure to act, which is an odd thing to warn against unless you have seen it happen.

Japan is still hard on failure

None of this adds up to "Japan tolerates failure." The opposite case is at least as strong, and Toyota supplies it.

In 2024, irregularities surfaced in Toyota's own vehicle certification testing. After cases at other companies in the group, the transport ministry ordered 85 firms to audit themselves; by the end of May, Toyota, Honda, Mazda, Suzuki and Yamaha Motor had all reported problems. At Toyota the count reached seven models in June and seven more in July, and on the 31st the ministry issued a corrective order requiring changes to the company's organisation.

That is the same corporate group, in the same year, where the CTO credits failure for speed. The contradiction dissolves once you notice that the word means two different things. A prototype that fails teaches you something nobody knew. A certification procedure that fails teaches you nothing and stops the trucks.

The gap widens further away from the big names. When smaller Japanese firms borrow, the owner has often personally guaranteed the company's debt, so a business failure follows the individual home. Japan's Small and Medium Enterprise Agency says plainly on its own site that this practice is pointed to as a factor discouraging bold business moves and early restructuring. A guideline to loosen it has existed since 2013 and a government reform programme since 2022, which tells you both that the problem is recognised and that it has not been solved.

The real story is narrower than a national trait. A handful of Japanese manufacturers have concluded that their development function had lost the ability to be wrong, and are buying it back.

Speed, buildings, paperwork

The three are not doing the same thing. Toyota is buying attempts, Murata bought floor space, MTC uses procedure.

Speed, buildings, paperwork, at three price points and resting on one admission: in a well-run factory, failure does not happen by default. Efficiency removes it. If you want it back, you have to put it in the plan and pay for it.

Your employer almost certainly has a training budget and an equipment budget. Does it have a line for being wrong? And if a project of yours collapsed next quarter, would anyone in the building treat that as information, or just as your fault?

References