Insights from the 2025 TokyoDev Developer Survey

Each year, TokyoDev runs a survey to better understand the experiences of international software developers in Japan. This year 989 people participated, up 21% from last year’s survey. These respondents answered 47 questions on topics such as compensation, working conditions, and AI. Following are some of our most important and surprising findings from the 2025 survey.

Median compensation was 9.5 million yen

Respondents had a median compensation of 9.5 million yen, up 1 million yen from 2024. 57% of participants reported that their compensation had increased since last year. Only 6% reported that it had decreased.

Compensation correlated highly with professional experience. Respondents with less than one year of experience made a median of 2.8 million yen, while those with 20+ years of experience made a median of 14.2 million yen.

International companies still pay better, but the gap is narrowing

Respondents working at an international company without operations in Japan earned the most, with a median salary of 13.5 million yen. Next came respondents working at a Japanese subsidiary of an international company at 11.5 million yen, followed by companies headquartered in Japan at 8.5 million yen.

Since we first asked about employer type in our 2020 survey, the pay at Japanese companies has always been lower compared to that at international businesses. This gap peaked in 2022, with respondents working at Japanese subsidiaries earning 93% more than those at Japanese-headquartered companies. Since then, the gap has narrowed every successive year: 73% in 2023, 47% in 2024, and finally 35% this year.

During this period, the compensation of respondents at Japanese-headquartered companies only rose slightly, from 7.5 to 8.5 million yen. This means that the reduced salary gap is mostly the result of falling compensation rates at international subsidiaries.

Japanese companies that adopt English have better practices

Japanese companies have told us that one advantage of hiring English speakers is that they are more up to date with the latest technologies and practices. Their belief was backed up by our survey results, which showed that a greater use of English at Japanese-headquartered companies correlated with a higher rate of adoption of software engineering best practices.

For example, 70% of respondents who used English frequently or always said their employer used continuous integration, compared to 45% of those who rarely or never used English. Similarly, 63% of frequent English users worked at organizations with monitoring and logging, compared to 34% of infrequent users. For automated testing, the figures were 59% versus 28%, and for infrastructure as code, 49% versus 25%. The only practice more common among infrequent English users was waterfall development, used by 30% compared to 15% of frequent users.

Greater English usage correlated with higher compensation

The more respondents used English, the higher their median compensation. Respondents who never used English had a median compensation of 5.5 million yen, whereas those who used it exclusively had a median of 10.5 million yen. This likely ties back to the earlier correlation between English usage and industry-standard best practices. Companies that adopt these best practices tend to prioritize hiring highly skilled developers, and are therefore more willing to offer competitive compensation.

These numbers partially reflect differences in professional experience. Those who never used English had a median of 4 years experience, whereas those who always used it had a median of 8 years.

However, when comparing respondents with similar experience levels, those who used English more enjoyed better compensation than those who used it less.

Experienced respondents used English more frequently

Respondents with greater professional experience were more likely to use English exclusively. Of those with 10+ years of experience, 74% used English only in the workplace, compared to 51% of those with 3 years or less.

As previously shown, the most attractive developer jobs in Japan—both in compensation and in developer culture—are typically found in English-speaking environments. Because these companies can draw from a broad international talent pool, they can afford to be selective and only hire more experienced developers.

In contrast, there are almost no junior-friendly roles in Japan that require no Japanese. As a result, the junior developers who do manage to find work here usually do so in positions that involve using Japanese.

Optional remote work decreased, but those who work remote were happier

In a continuation of last year’s trend, more companies moved away from optional remote work and embraced hybrid work schedules. In 2024, 38% of respondents could decide for themselves if they wanted to work remotely; in 2025, that number fell to 32%. Hybrid work schedules, by contrast, rose from 37% to 43% this year.

Respondents at organizations that permit remote work were happier with their job. On a scale of 1-5, those who could work fully remotely had an average happiness score of 3.8, compared to 2.9 for respondents who could never work remotely. Those in hybrid work environments scored an average of 3.4.

Not only were respondents who had to work from an office less happy, but they were also more likely to be looking for a job: 42% of them were actively job-hunting, compared to only 20% of those in fully remote environments.

The negativity associated with in-office mandates may be less about the value of remote work in and of itself, and more that such companies tend to be worse places to work overall. For instance, when respondents could choose whether to work fully remotely or not, 45% of respondents still chose to go into the office regularly, showing that a large percentage of respondents would still prefer to regularly work from an office. Furthermore, those who did go in were also less likely to be looking for a new job, at a rate of 12%, compared to 17% of those who worked regularly from a dedicated home office, showing there actually may be some retention benefits with employees coming into the office.

Views on life in Japan were shaped by nationality

Residents of Japan, regardless of nationality, considered “convenience” one of Japan’s greatest benefits.

However, some aspects of Japan were rated highly by certain nationalities and not others. For instance, 38% of Americans and 47% of Japanese rated healthcare as being one of the top positive aspects of living here, but for people of other nationalities, only 16% indicated that Japanese healthcare was a major benefit.

The most negatively-rated aspect of life in Japan was the same across all nationalities: poor compensation. After that, non-Japanese citizens most often cited isolation and cultural barriers as problematic, while Japanese citizens were more likely to mention social pressure to conform.

AI usage was widespread, but dropped with experience

When we asked who used AI tools, 94% of respondents said they did regularly. The most popular use cases were code generation, learning and research, and translation.

Many used it very regularly: 33% of respondents generated at least half the code they produced through AI, and 38% of respondents were using it daily.

More experienced respondents used AI less frequently. The median experience was 10 years for those who never used it, compared to 6 years for those who used it constantly. It’s possible that the responsibilities of more experienced respondents are in general less suited to AI, as those who strongly disagreed that AI had made them more productive had a median of 8.5 years of experience, compared to 6 years for those who strongly agreed.

The gender gap widened with experience

Of our respondents, 87% were men and 12% were women. Because only 1% of respondents didn’t fit into those categories, we didn’t have enough data to perform further analyses on them without compromising anonymity. If future surveys have a larger number of non-binary respondents, we will be able to include them in analyses while still protecting their privacy.

In some areas, we found little variation between the responses from men and women. For instance, there was no significant difference in their employment status, their happiness with their job, whether they worked for a Japanese-headquartered company or not, the industry they worked in, or the size of their employer.

In other areas, however, there were noteworthy differences. For instance, women respondents tended to be less experienced than men, with a median of 5.5 years of professional experience compared to 8 years for men. They were more likely to live in Tokyo: 69% of them lived there compared to 58% of men. They were more likely to be coding bootcamp grads as well, at a rate of 21% compared to 10% for men.

Women were also significantly more concerned about AI’s broader ramifications, with 36% of women citing ethical considerations compared to 19% of men, 32% worried about privacy compared to 21% of men, and 28% concerned about its impact on the environment compared to 18% of men.

Women respondents were far more likely to have experienced gender-based discrimination in the workplace, with 32% of women reporting it compared to 2% of men.

Overall, the median compensation of men respondents was 46% higher than that of women respondents. However, when analyzing respondents by experience level and gender, women who had 6 or fewer years of experience actually had a median compensation that was the same as men. For more experienced women, though, the gap was considerable. Among those with 7–9 years experience, men earned 62% more than women. For those with 10 or more years of experience, men earned 67% more.

The low representation of women on engineering teams creates a range of challenges. Small minorities are more easily left out of informal networks and after-work activities, and in homogeneous teams, stereotypes can persist simply because there are no counterexamples. Male-dominated teams can also cause candidates to self-select out, further perpetuating the imbalance and making it harder to change. Addressing this requires a consistent and collective effort. If you'd like to help, see a four-stage approach for hiring women on your engineering team.

Plus so much more

There are many more insights to be found in the full survey results. We also offer the ability to create your own custom charts, allowing you to explore the data any way you like. So please, dive in and take a look. If you find any other interesting connections, you’re welcome to share them with our community!