OpenAI continues to attract tons of venture capital, its accumulated take reportedly hovering at $6.5 billion most recently. And a valuation that now exceeds $150 billion might appear to give the Microsoft and Softbank-backed startup an insurmountable advantage in the generative artificial intelligence (GenAI) space.
But emerging players like Paris-based Mistral AI and Tokyo-based Sakana AI are gaining ground, says entrepreneur Ilya Kulyatin, who founded Tokyo AI, a community of over 1,000 AI professionals. Mistral’s focus on smaller, efficient models presents a challenge to OpenAI’s large, resource heavy GPT models, Kulyatin argues.
Japan is taking a similar approach to cultivate a strong AI push. Sakana AI is positioning itself as a key player by focusing on resource-efficient technologies. In February, it received a government grant to enhance its access to supercomputers and GPUs, which are used to train foundational models.
With $200 million in Series A funding and a strategic partnership with Nvidia, Sakana AI is demonstrating the growing importance of smaller AI models, particularly in resource-constrained environments like Japan that are considered GPU-poor. Sakana AI’s work in military and drone-related technologies, backed by the Japanese government, exemplifies this focus on smaller, more efficient AI models.
The Japanese effort faces significant headwinds, however. The venture capital environment in Tokyo remains conservative, often focusing on smaller exits rather than scaling startups to compete globally, Kulyatin notes. This cautious approach makes luring foreign investment challenging. Japan also lags in AI talent development, he adds, as local universities are unable to compete with institutions like Stanford or Carnegie Mellon universities. To close the talent gap and become more competitive, Japan looks to attract top-tier talent—not to mention investors—from abroad.