This is Ryuta Hamamoto from TIMEWELL.
On March 17, 2026, NVIDIA CEO Jensen Huang unveiled a list of 103 "AI-native" companies during his GTC 2026 keynote. The list spans everything from OpenAI and Anthropic to autonomous driving, drug-discovery AI, and humanoid robotics — yet not a single Japanese company made the cut. In this article, I'll walk through the companies in each category and examine the structural reasons behind Japan's absence.
Jensen Huang's "New Industrial Map"
The GTC 2026 keynote was far more than a product announcement. Huang laid out a new industrial architecture built around the concept of the "AI factory." Data centers are no longer places to store files and run software that humans operate by hand. They are becoming environments where AI agents autonomously execute workflows and route context to the right destinations. Huang noted that compute demand for NVIDIA GPUs had increased "a million times over the past several years," and revealed that the company expects cumulative revenue of $1 trillion between 2025 and 2027.
It was in this context that the 103 AI-native companies appeared. According to Forbes reporting, the slide organized them into 12 categories: AI for Auto, AI for Customer Support, AI for Engineering, AI for Healthcare, AI for Robotics, AI for Search, and more — illustrating how AI-native companies are emerging across every industry. Huang himself joked during the keynote: "I went back and forth on whether to show more companies or fewer, and I settled on making the text so small that no one can read it. So no one's feelings get hurt." But the fact that no Japanese company appeared on that slide carries a weight that goes beyond humor [1].
Foundation Models — The "Brains" Driving Global AI
Among the 103 companies, the most high-profile group is the cluster of firms developing large language models (LLMs) and foundation models. OpenAI (US, founded 2015) needs no introduction — its GPT series set off the generative AI boom. Anthropic (US, founded 2021), founded by former OpenAI researchers, takes a safety-first approach to AI development through what it calls "Constitutional AI." Its agentic system, Claude Code, has already become part of many engineers' daily workflows.
xAI (US, founded 2023), founded by Elon Musk, runs the Grok model on X (formerly Twitter) and made headlines for training with 100,000 H100 GPUs. France's Mistral AI (founded 2023) also made the list, founded by former Google DeepMind researchers including Arthur Mensch. The company has earned strong support from European enterprises with models designed for EU data sovereignty requirements, multilingual support, and on-premises deployment options. Mistral AI has also joined NVIDIA's newly launched Nemotron Coalition and is co-developing the next-generation foundation model, Nemotron 4 [2].
Canada's Cohere (founded 2019) focuses on enterprise LLMs, while Perplexity (US, founded 2022) has grown rapidly as an AI-native search engine. NVIDIA has made direct investments in Perplexity, and the relationship between the two companies has evolved into a partnership that goes beyond a simple customer relationship. In this space, US companies dominate — though France's Mistral and Canada's Cohere stand out. No Japanese company building its own foundation model for the global stage appeared on the slide [3].
Interested in leveraging AI?
Download our service materials. Feel free to reach out for a consultation.
Autonomous Driving and Physical AI — The Rise of Chinese Players
The "AI for Auto" category features companies working on autonomous driving. Waymo (US, subsidiary of Alphabet) is already running commercial robotaxi services in San Francisco and Phoenix, making it one of the defining AI-native companies in this space. Aurora (US, founded 2017) focuses on autonomous trucking and went public via SPAC in 2021, with its Aurora Driver platform targeting commercial logistics automation.
Notably, this category includes Chinese companies as well. Pony.ai (founded 2016) became the first company to receive a permit for fully driverless robotaxi operations at city scale in China, with commercial operations underway in Shenzhen. WeRide (founded 2017) is another Chinese autonomous driving startup developing robotaxis and self-driving buses. US-based Nuro (founded 2016) specializes in autonomous delivery robots, aiming to transform last-mile logistics.
In the physical AI space — AI that operates in the physical world — humanoid robotics companies are particularly prominent. Figure AI (US, founded 2022) received direct investment from NVIDIA and reached a $39 billion valuation in its Series C round in September 2025. Physical Intelligence (US, founded 2024) is developing generalist control AI for robots, while 1X Technologies (Norway, founded 2014) is known for its humanoid robot "NEO." Agility Robotics (US, founded 2015) has deployed its bipedal robot "Digit" for warehouse operations. FieldAI (US) specializes in industrial AI agents and is listed as a collaborating company in NVIDIA's physical AI data factory blueprint [4].
Japan's Toyota and Honda have adopted NVIDIA's DRIVE AGX platform, but under the "AI-native" definition — companies founded with AI as their core — Japan's autonomous driving startups didn't make the list, despite the existence of companies like Tier IV and Turing.
Coding and Engineering — Companies Fundamentally Changing How Developers Work
AI-native companies in the coding and engineering space were also a major part of the list. Cursor (US) is an AI-powered code editor that Jensen Huang personally called out by name as one of six tools used by "every engineer at NVIDIA." It's not simply an autocomplete tool — its ability to understand an entire codebase and act as an AI agent that proposes and implements changes has earned it high marks in the developer community.
Cognition (US, founded 2023) created Devin, billed as the world's first AI software engineer, capable of understanding a full repository, fixing bugs, and implementing new features autonomously. Harvey (US, founded 2022) focuses on legal AI, automating contract review and legal research — and Huang mentioned it as one of the six AI tools he personally uses. Replit (US, founded 2016) is a browser-based AI coding platform that handles code generation, execution, and deployment end to end. Lovable (Sweden) is an AI app-building platform where users can generate applications from prompts alone. OpenEvidence (US) is a medical-evidence AI that Huang also cited as a personal-use tool [5].
This category appears to have included multiple competitors to GitHub Copilot, including Codium (now Qodo, founded in Israel), making it clear that the transformation of software development through AI is a central theme within NVIDIA's ecosystem.
Healthcare and Life Sciences — The Frontier of Drug-Discovery AI
According to an analysis by LinkedIn's Healthcare AI Guy, at least eight companies in the healthcare space were confirmed on the list. Arc Institute (US, founded 2021) is a nonprofit research organization bringing together researchers from Stanford, UC Berkeley, and UCSF to advance bioAI. Biohub (US), part of the Chan Zuckerberg Initiative founded by Mark Zuckerberg and Priscilla Chan, uses AI to map infectious diseases and create atlases of human cells.
Isomorphic Labs (UK, founded 2021) spun out of Alphabet's DeepMind and uses AlphaFold technology for molecular-level prediction in drug discovery — its UK base offers a notable counterpoint to the US-centric nature of much of the list. Chai Discovery (US, founded 2023) develops AI drug-discovery models specializing in predicting molecular interactions. The Institute for Protein Design at the University of Washington also appears on the list — led by Professor David Baker, the institute won the Nobel Prize in Chemistry in 2024 [6].
Recursion Pharmaceuticals (US, founded 2013) has a deep relationship with NVIDIA and runs a drug-discovery platform that combines proprietary lab-generated data with AI. Japan has its own efforts in AI drug discovery — including Preferred Networks, MOLCURE, Elix, and projects at RIKEN — but none of them made it onto the 103-company slide.
AI Infrastructure and Cloud — The Backbone of the "AI Factory"
The companies most directly enabling NVIDIA's "AI factory" vision are those in the AI infrastructure space. CoreWeave (US, founded 2017) is a Kubernetes-native AI cloud platform built specifically around NVIDIA GPUs, known for fast instance provisioning. It went public in 2025 and has been expanding rapidly, including the acquisition of Weights & Biases (W&B). Nebius (originally Russian, now headquartered in the Netherlands, reorganized 2024) is a full-stack AI cloud that grew out of Yandex's cloud division and announced a major partnership with NVIDIA at GTC 2026.
Lambda (US, founded 2012) is a veteran GPU cloud provider, offering NVIDIA GPU access at accessible price points for researchers and startups. Together AI (US, founded 2022) is a platform focused on training and serving open-source models. Cerebras (US, founded 2016) has attracted attention for its wafer-scale AI chip technology, while Groq (US, founded 2016) developed the LPU (Language Processing Unit), an inference-optimized chip. NVIDIA actually announced an inference server rack at GTC 2026 combining Groq's silicon with its own GPUs — a show of flexibility in which even a competitor becomes a collaborator.
Scale AI (US, founded 2016) holds a commanding share of the AI training-data labeling and quality assurance market and is known for contracts with the US government and military. Databricks (US, founded 2013) and Snowflake (US, founded 2012) are the two giants of the data platform world, both rapidly building out their AI workload capabilities [7].
Enterprise AI and Agents — AI Permeating the Workplace
In the enterprise space, Glean (US, founded 2019) has established itself as an enterprise search AI. Its platform searches across internal documents, Slack, email, and knowledge bases and uses AI to generate answers. Hebbia (US, founded 2020) provides document analysis AI for finance and legal professionals, capable of instantly understanding and summarizing hundreds of pages of contracts or financial filings. Decagon (US) builds AI-powered customer support agents.
LangChain (US, founded 2022) has become the de facto standard framework for building LLM applications and has joined the Nemotron Coalition alongside Cursor, Mistral AI, and Perplexity [8].
ZEROCK, offered by TIMEWELL, is also operating in the enterprise AI space, enabling internal information search using GraphRAG and knowledge control. For Japanese companies to break into NVIDIA's list, gaining global recognition in enterprise AI may well be one of the most viable paths forward.
Creative AI and Players from Emerging Markets
Media and creative AI companies are also a meaningful part of the 103. Runway (US, founded 2018) is a pioneer in AI video generation — its Gen-3 model is beginning to appear in Hollywood productions. Pika Labs (US, founded 2023), founded by Stanford researchers, is a video-generation AI that has raised $55 million. Black Forest Labs (Germany, founded 2024), founded by former Stable Diffusion developers, has made waves with its FLUX image-generation model. ElevenLabs (US/UK, founded 2022) is a leader in AI voice synthesis.
Companies from emerging markets also deserve attention. Sarvam AI (India, founded 2023) is building India's sovereign language AI with models covering Hindi and multiple other Indian languages. Thinking Machines Lab (Philippines) is also among the Nemotron Coalition members.
The list includes companies from Israel, Norway, Germany, the Philippines, and India, among others. The global distribution of AI-native companies is no longer a Silicon Valley story alone. That makes Japan's complete absence all the more striking — it suggests that the issue is not simply one of technical capability, but of not being visible within NVIDIA's ecosystem [9].
Why Zero Japanese Companies — A Structural Problem
Honestly, the absence of Japanese companies from this list didn't shock me, even if it did surprise me. There are several reasons.
First, the definition of "AI-native" doesn't align well with Japan's AI ecosystem. When Jensen Huang says AI-native, he means companies that put AI at their core from day one — companies that couldn't exist without it. Japan has excellent AI companies like Preferred Networks, Sakana AI, and AI inside, but by the criteria of how much NVIDIA GPU compute they consume at scale and how much global presence they command, they haven't yet broken into the top 103.
Second, there are structural constraints in Japan's startup ecosystem. Of the roughly $150 billion in annual VC investment flowing into AI-native companies globally, Japan's share is minimal. In a world where Scale AI is valued at multiple billions of dollars and Figure AI raises over $1 billion in a single round, Japanese AI startups are raising one or two orders of magnitude less.
A third factor that can't be ignored is the depth of relationships with NVIDIA. Many of the 103 companies are direct NVIDIA portfolio investments and primary users of NVIDIA's DGX Cloud, NIM, Omniverse, and DRIVE AGX platforms. NVIDIA held an "AI Summit Japan" in Tokyo in November 2024, where Jensen Huang appeared in conversation with SoftBank's Masayoshi Son. But those engagements are about large enterprises building AI infrastructure — a different story from producing AI-native companies from scratch [10].
The Industrial Map of the Next Ten Years
It's tempting to dismiss this slide as "NVIDIA's favorites list," but I don't see it that way. The 103 companies on this list have the potential to fundamentally rewrite existing industries over the next decade. Waymo will drive on public roads. Harvey will automate legal work. Isomorphic Labs will accelerate drug discovery. Figure AI will send humanoid robots into factories.
Many of the companies on the list are still unprofitable. But in Huang's view, this probably looks exactly like the late 1990s at the dawn of the internet. NVIDIA's recognition of 103 companies as "AI-native" isn't a guarantee of profitability — it's a declaration that the platform shift is real.
And the fact that Japan isn't named in that declaration should be a signal worth taking seriously for Japan's AI industry. The time has come to think carefully about where the global AI industry's main battleground lies, and how Japan positions itself there. It's not enough to buy NVIDIA GPUs. Japan needs to produce companies that make NVIDIA want to include them on the next slide.
References
[1] Forbes. "A List Of All 103 AI Native Companies Nvidia's Jensen Huang Presented." https://www.forbes.com/sites/josipamajic/2026/03/17/a-list-of-all-103-ai-native-companies-nvidias-jensen-huang-presented/ (2026-03-17)
[2] Constellation Research. "Nvidia GTC 2026: We're a software company too." https://www.constellationr.com/insights/news/nvidia-gtc-2026-were-software-company-too (2026-03-17)
[3] Investing.com. "NVIDIA at GTC 2026: AI Expansion and Strategic Partnerships (Transcript)." https://www.investing.com/news/transcripts/nvidia-at-gtc-2026-ai-expansion-and-strategic-partnerships-93CH-4564073 (2026-03-17)
[4] NVIDIA Newsroom. "NVIDIA Announces Open Physical AI Data Factory Blueprint." http://nvidianews.nvidia.com/news/nvidia-announces-open-physical-ai-data-factory-blueprint-to-accelerate-robotics-vision-ai-agents-and-autonomous-vehicle-development (2026-03-17)
[5] Yahoo Finance. "Jensen Huang name-checks 6 AI companies." https://finance.yahoo.com/news/jensen-huang-name-checks-6-085752539.html (2026-03-17)
[6] LinkedIn (Healthcare AI Guy). "Jensen named 103 companies 'AI Native' at NVIDIA GTC." https://www.linkedin.com/posts/healthcareaiguy_jensen-named-103-companies-ai-native-at-activity-7439794964995223553-JNmB (2026-03-17)
[7] LinkedIn (Deedy Das). "Every single one of the 103 companies Jensen called AI Native at NVIDIA GTC today." https://www.linkedin.com/posts/debarghyadas_every-single-one-of-the-103-companies-jensen-activity-7439529024294588416-m_2g (2026-03-17)
[8] NVIDIA Newsroom. "NVIDIA Launches Nemotron Coalition of Leading Global AI Labs." http://nvidianews.nvidia.com/news/nvidia-launches-nemotron-coalition-of-leading-global-ai-labs-to-advance-open-frontier-models (2026-03-17)
[9] note.com. "NVIDIAのジェンセン・ファンCEOが発表した『103社のAIネイティブ企業』のリストに日系企業が一つもない." https://note.com/fair_godwit545/n/n067b4d6c3041 (2026-03-17)
[10] Fortune. "The day tech stood still to watch the Jensen Huang show." https://fortune.com/2026/03/17/the-day-tech-stood-still-to-watch-the-jensen-huang-show-nvidia/ (2026-03-17)