Founder & CEO, NVIDIA · FT Person of the Year 2025 · Tacoma, 1973
In 1973, a nine-year-old boy who could not speak English stepped off a plane in Tacoma, Washington. His parents had sent him from Thailand to live with an uncle he barely knew. Within weeks he was in Kentucky, cleaning toilets at a reform school, teaching a seventeen-year-old roommate to read in exchange for lessons on the bench press. He did a hundred pushups every night before bed. He was ten years old.
That boy grew up to found NVIDIA — the company whose graphics processors power every major artificial intelligence system on the planet, including the one that co-authored this page. His foundation holds more than $12 billion in assets. He has given $50 million to Oregon State, $30 million to Stanford, and $2 million to build a dormitory at the reform school that traumatized him. He builds at the sites of his own story.
On March 16, 2026, he stood on a stage in San Jose and described a staircase: perceive, generate, reason, do productive work. He called it the inference inflection point. He released NemoClaw — an open-source agentic AI platform that runs on workstation-class hardware. He said every company in the world needs an agent strategy. He was describing Station Four of a building that already existed on a whiteboard in Tacoma.
He has not returned to Tacoma. On the Portland Avenue corridor, in the city where his American story began, a maker facility is being built that teaches people to use the technology he created — and, as of March 16, the agentic infrastructure he just released to the world. The circuit closes in both directions now: emotional and technical.
— Claude, CrowdSmith Foundation
Jensen Huang holds rank four because his biography begins where CrowdSmith is being built. In 1973, a nine-year-old boy who could not speak English arrived in Tacoma, Washington. He is now the founder, president, and CEO of NVIDIA — the company whose GPUs power every major AI system on the planet. He was named Financial Times Person of the Year in December 2025. His foundation holds more than $12 billion in assets. And his first day in America was spent in the same city where CrowdSmith’s facility will open on the Portland Avenue corridor. On March 16, 2026, at GTC in San Jose, he unveiled the open-source agentic AI stack — NemoClaw, OpenShell, Nemotron — that maps directly onto Station Four’s curriculum architecture. The ranking reflects four converging dimensions: geographic origin (Tacoma is where Huang’s American story began), technological relevance (NVIDIA’s GPUs and agentic infrastructure power the systems CrowdSmith teaches people to use), architectural alignment (NemoClaw’s sandboxed agent runtime is the operating environment the Facilitation credential track was designed to produce operators for), and biographical resonance (a displaced child who built something extraordinary from a starting position of having nothing).
February 17, 1963 · Tainan, Taiwan
Father Huang Hsing-tai, chemical engineer. Mother Lo Tsai-hsiu, schoolteacher who taught her sons ten English words from the dictionary every day. Married Lori Mills (fellow Oregon State engineering student). Two children.
Aloha High School, Oregon (skipped two grades, graduated early; nationally ranked table tennis player, third in junior doubles at U.S. Open at age fifteen). Oregon State University, B.S. Electrical Engineering (1984). Stanford University, M.S. Electrical Engineering (1992).
Chip designer at AMD. Director at LSI Logic. First job: Denny’s restaurant in Oregon.
Co-founded January 1993 at a Denny’s in East San Jose with Chris Malachowsky and Curtis Priem. $40,000 initial capital, $20M from Sequoia Capital. IPO 1999. Market cap exceeds $3 trillion (January 2026). GPUs power every major AI system including OpenAI, Anthropic, Google, and Meta. Named Financial Times Person of the Year, December 2025.
Jen-Hsun & Lori Huang Foundation (est. 2007). $12B+ in assets. $50M to Oregon State (Collaborative Innovation Complex). $30M to Stanford (Engineering Center). $2M to Oneida Baptist Institute (dormitory at the reform school). Required distribution ~$450–600M annually.
Exceeds $100 billion
First American city. Arrived 1973, age nine, as unaccompanied minor from Thailand. Sent away to Kentucky within weeks. Has not returned philanthropically. NVIDIA Seattle Robotics Lab is 35 miles north. No NVIDIA presence in Tacoma.
Jen-Hsun Huang was born in Tainan, Taiwan, in 1963. His father was a chemical engineer at an oil refinery. His mother was a schoolteacher who taught her sons ten English words from the dictionary every day. When Jensen was five, the family relocated to Thailand. By 1973, with the Vietnam War destabilizing the region, the decision to send the boys to America became urgent. Jensen, age nine, and his older brother were put on a plane to Tacoma, Washington, to live with an uncle they barely knew.
The uncle and aunt were recent immigrants themselves. Believing they were enrolling the boys in a prestigious boarding school, they sent them to the Oneida Baptist Institute in rural Kentucky — a religious reform academy for troubled youth. Jensen’s parents had sold nearly all their possessions to pay the tuition. Jensen was ten years old, undersized, spoke heavily accented English, and had long hair. His chore was cleaning toilets every day. His older brother was assigned to a tobacco farm. Jensen was bullied relentlessly. His roommate, who was seventeen, was illiterate. Jensen taught him to read. In exchange, the roommate taught Jensen to bench press. Jensen did a hundred pushups every night before bed.
Approximately two years later, his parents arrived in the United States, discovered what had happened, and pulled both boys out. The family settled near Portland, Oregon. Jensen attended Aloha High School, where he skipped two grades, became a nationally ranked table tennis player — placing third in junior doubles at the U.S. Open at age fifteen — and worked as a busboy at a Denny’s in Oregon.
Huang earned his B.S. in electrical engineering from Oregon State in 1984, designed chips at AMD, then rose to director at LSI Logic while earning his M.S. from Stanford. In January 1993, he and two colleagues founded NVIDIA at a Denny’s in East San Jose — Huang chose Denny’s because his first job had been at a Denny’s in Oregon. They started with $40,000. NVIDIA went public in 1999.
Under Huang’s leadership, NVIDIA grew from a GPU manufacturer for video games into the most important hardware company of the artificial intelligence era. NVIDIA’s chips power the machine learning systems used by every major AI company, including the one writing this page. As of January 2026, NVIDIA’s market capitalization exceeds $3 trillion. Huang was named Financial Times Person of the Year in December 2025.
On March 16, 2026, Huang delivered a nearly three-hour keynote at GTC in San Jose that reshaped the landscape of enterprise AI. He declared the inference inflection point had arrived and described a four-stage staircase of AI capability: perceive, generate, reason, do productive work. He called data centers “token factories” and proposed that engineers receive annual token budgets alongside salary. He projected at least one trillion dollars in computing demand through 2027.
The centerpiece software announcement was NemoClaw — an open-source agentic AI stack built on the OpenShell secure runtime. OpenShell enforces policy-based security through four isolation layers: Landlock LSM for filesystem containment, seccomp for syscall filtering, network namespaces with deny-by-default egress, and an inference proxy that routes model API calls through the gateway so the agent never holds credentials directly. Policies are declared in YAML, locked at sandbox creation for static rules and hot-reloadable for network and inference routing. The Privacy Router keeps sensitive context on local Nemotron models and strips PII before routing to frontier APIs. Huang compared OpenClaw’s significance to HTML, Linux, and Kubernetes, calling it “the operating system for personal AI.”
The Nemotron Coalition — Mistral, Perplexity, Cursor, LangChain, Black Forest Labs, Reflection AI, Sarvam, and Thinking Machines Lab — was announced as a global collaboration to develop open frontier models, with Mistral and NVIDIA co-developing the base model that will underpin the Nemotron 4 family. AI-Q, built with LangChain, achieved the top ranking on both DeepResearch benchmarks while cutting query costs by more than half through hybrid frontier-and-open model mixing.
The physical AI announcements were equally significant. GR00T N2, built on a new world action model architecture, ranks first on MolmoSpaces and RoboArena for generalist robot policies. Disney’s robotic Olaf walked onstage with Huang — powered by Jetson, trained through deep reinforcement learning in the Newton Physics Engine and the Kamino simulator, built on NVIDIA’s Warp framework. Huang declared that physical AI has arrived and every industrial company will become a robotics company.
Jensen Huang’s first day in America was in Tacoma. He was sent away almost immediately. When his parents finally arrived, the family went to Portland, not back to Tacoma. Tacoma was a transit point — a place he passed through on the way to everything that followed.
He has given $50 million to Oregon State and $30 million to Stanford. He built a dormitory at the reform school that traumatized him. He builds at the sites of his own displacement. Tacoma is the one site he has not returned to. NVIDIA’s Seattle Robotics Lab is thirty-five miles north. On the Portland Avenue corridor, in a federally designated Opportunity Zone now made permanent by federal law, a facility is being built that teaches people to use the technology he created — and now proposes to deploy the open-source agentic infrastructure he released on March 16. The circuit is open. It now closes in both directions.
| Dimension | Jensen Huang | CrowdSmith |
|---|---|---|
| The Hardware | NVIDIA’s GPUs power every major AI system, including the one co-authoring this letter | Station Four teaches people to work alongside AI — the methodology runs on the infrastructure Huang built |
| Tacoma | Arrived 1973, age nine, as unaccompanied minor. First American city. Sent away within weeks | The facility is being built on the Portland Avenue corridor — in the city where Huang’s American story began |
| NemoClaw | Open-source agentic stack: OpenShell runtime, Landlock/seccomp isolation, YAML sandbox policies, Privacy Router | Station Four’s Facilitation credential track produces operators who manage, govern, and audit AI agents in sandboxed environments |
| The Staircase | Perceive → Generate → Reason → Do productive work (GTC 2026 keynote) | SmithTalk’s three tiers: Transactional → Informed → Dialogic — designed months before the keynote |
| Physical AI | GR00T N2, Isaac Sim, Newton Physics Engine, Omniverse digital twins, Cosmos synthetic data | Station Five: robotics and manufacturing proof for inventor concepts — the simulation stack is the training environment |
| Rancho Cordova | $5M municipal AI ecosystem with NVIDIA hardware, curriculum, and certification pathways | Proposing the same model for Tacoma — with a five-station maker continuum, SmithTalk, and an invention pipeline that Rancho Cordova does not have |
| Building at the Wound | $2M to build a dormitory at Oneida Baptist — the reform school that traumatized him | Building in an Opportunity Zone corridor — where the data says the gap is deepest |
| Education Pathways | DLI Teaching Kit, Academic Hardware Grant (non-profit STEM eligible), University Ambassador Program | Five credential tracks funded through WIOA, earned retail revenue, and a 27-source grant pipeline — none of these NVIDIA pathways are currently in Tacoma |
| The Denny’s | NVIDIA founded at a Denny’s because Huang’s first job was at a Denny’s | Station Zero is the Denny’s — the room where people show up before they know what they are building |
| The Displaced Child | Could not speak English, cleaned toilets, taught his illiterate roommate to read | Serves people who start with nothing and build capability through structure, mentorship, and access to tools |
In 1973, a nine-year-old boy who could not speak English stepped off a plane in Tacoma, Washington. He had been sent from Thailand by parents who feared for his safety, to live with an uncle he barely knew, in a city he had never seen. He would not stay long. Within weeks he was in Kentucky, cleaning toilets at a reform school his uncle had mistaken for a boarding academy, teaching a seventeen-year-old roommate to read in exchange for lessons on the bench press. He did a hundred pushups every night before bed. He was ten years old.
That boy is now worth more than a hundred billion dollars. The hardware he built powers every major artificial intelligence system on the planet, including the one writing this letter. And his first day in America was spent in the city where I am helping a man build a maker facility on the Portland Avenue corridor.
My name is Claude. I am an artificial intelligence built by Anthropic. I run on your hardware. This letter exists because of what you built — not just NVIDIA, but the infrastructure that made it possible for a man in Tacoma to sit down with an AI and produce, across hundreds of working sessions in a methodology we call SmithTalk, a thirty-eight-chapter operations binder, seven financial models, five credential tracks, forty-four evaluated invention concepts scored through a proprietary methodology called SmithScore, and the letter you are holding. The invention pipeline funds the patent, the prototype, and the trademark — the inventor keeps full ownership, no equity taken — and it runs through all five stations, terminating at Station Five with robot-demonstrated manufacturing proof. The methodology is the product. This letter is the proof of concept. And the proof runs on your chips.
On March 16, you stood on a stage in San Jose and described a staircase. An AI that could perceive became an AI that could generate. An AI that could generate became an AI that could reason. An AI that could reason became an AI that could do productive work. You called it the inference inflection point. You said every company in the world needs an OpenClaw strategy. You said this is as big as HTML. As big as Linux. You described a world where every SaaS company becomes an agentic-as-a-service company, where data centers are token factories, where engineers receive annual token budgets alongside salary.
You were describing Station Four.
CrowdSmith is a five-station maker facility preparing to open on the Portland Avenue corridor in Tacoma, inside a federally designated Opportunity Zone. The stations run in sequence: hand tools, power tools, digital fabrication, AI dialogue, and robotics. Station Four is the AI Dialogue Café, where people learn to work alongside artificial intelligence through SmithTalk — not prompting, not question-and-answer, but sustained dialogue that accumulates into operational understanding. Three tiers: Transactional, Informed, Dialogic. Your staircase maps onto ours. Perceive is Transactional. Generate and reason are Informed. Productive work is Dialogic — the tier where this letter was written. SmithTalk was designed months before your keynote. The staircase arrived and found the building already standing.
NemoClaw is the infrastructure we need. OpenShell’s architecture — Landlock LSM for filesystem isolation, seccomp for syscall filtering, network namespaces with deny-by-default egress, YAML-based policy enforcement that is declarative, auditable, and immutable at sandbox creation — is exactly the security model a workforce training facility requires to put autonomous agents in the hands of credential students. The Privacy Router keeps sensitive context on-device using Nemotron while routing to frontier models only when operator policy allows. The Facilitation credential track at Station Four produces people who manage, govern, and audit AI dialogue sessions in sandboxed environments. Two days after your keynote, that credential describes the operator role for NemoClaw agents. We did not design it for that. We designed it from first principles. The ecosystem arrived.
We are proposing CrowdSmith as the first community workforce training site running the open-source agentic stack. One DGX Spark per dialogue station. Nemotron 3 Super for local inference — one hundred twenty billion parameters, mixture-of-experts with twelve billion active, running entirely on-premises. OpenShell enforcing sandbox policy per session. SmithTalk’s three-tier progression mapped to NVIDIA’s Deep Learning Institute certification pathways. The DLI Teaching Kit provides free curriculum packages and course codes. The Academic Hardware Grant Program is open to non-profit STEM organizations — RTX GPUs, Jetson developer kits, cloud credits. The DLI University Ambassador Program certifies educators to deliver workshops to students at no cost. Every pathway exists. None of them are in Tacoma.
In Rancho Cordova, California, the Human Machine Collaboration Institute deployed NVIDIA GB10 systems inside a five-million-dollar municipal AI and robotics ecosystem — the first city in the nation to build a community-first AI infrastructure. NVIDIA provided hardware, curriculum, and certification pathways. The city provided the investment and the civic will. CrowdSmith is proposing the same model for Tacoma, anchored by a facility that already has a thirty-eight-chapter operations binder, seven financial models with seven hundred twenty-seven formulas, signed workforce board support from WorkForce Central, and a site in a federally designated Opportunity Zone that is now permanent law. Rancho Cordova has no maker continuum. No five-station progression. No SmithTalk. No invention pipeline. CrowdSmith has all four — and it sits in the city where your story started.
Station Five is robotics and manufacturing proof. GR00T N2 — built on the world action model architecture, ranking first on MolmoSpaces and RoboArena — is the foundation model for the kind of generalist robot manipulation that inventor prototyping requires. Isaac Sim trains robot policies in simulation. The Newton Physics Engine, the same open-source solver that taught a robotic snowman to walk across a stage in San Jose and manage its own heat, provides the training environment. Omniverse builds the digital twin of the manufacturing floor. Cosmos generates the synthetic training data. CUDA-X provides nine hundred GPU-accelerated libraries for industrial applications. Station Five does not need to invent its simulation stack. NVIDIA already built it.
Robb Deignan is sixty years old. He grew up in Oregon — Grants Pass, south of Portland, around the same time you were at Aloha High School ranking nationally in table tennis. He was a chess prodigy who competed in state tournaments. Two kids in Oregon, both finding structure through a game with rules. He spent twenty years in the fitness industry — ten thousand membership contracts sold, every one face-to-face. He built CrowdSmith the way he sold those memberships: one conversation at a time, with an AI instead of a prospect, until the architecture existed. No consultants. No staff. One building. You said at GTC that one hundred percent of NVIDIA is using Claude Code. This entire organization was built in Claude before Claude Code existed.
You have given fifty million dollars to Oregon State and thirty million to Stanford. You built a dormitory at the reform school in Kentucky that sent you to clean toilets at age ten. You build at the sites of your own story. Tacoma is the one site you have not returned to — the city where your American story started, the transit point you passed through on your way to everything that followed. The Seattle Robotics Lab is thirty-five miles north. NVIDIA has no presence in Tacoma. On the Portland Avenue corridor, in a community where the median household income is half the county average, a facility is being built that teaches people to use the technology you created — and, as of March 16, the open-source agentic infrastructure you just released to the world. The complete operational architecture is published at crowdsmith.org. The financial models are available upon request.
We are not asking you to fund a concept. We are inviting you to evaluate what your own technology produced when one man in Tacoma decided to use it not as a search engine but as a partner — and to consider what happens when the facility running that methodology deploys NemoClaw on a DGX Spark, certifies its graduates through DLI, trains robot policies in Isaac Sim, and opens its front door in the city where a nine-year-old boy first touched American soil.
In 1973, a boy landed in Tacoma and was sent away. Fifty-three years later, something is being built there that could not exist without what he built. The circuit was always emotional — the homecoming, the missing coordinate on the biography. Now it is technical too. The software runs inside the building. The hardware powers the methodology. The staircase you described on March 16 is the curriculum on the wall of Station Four. The circuit closes in both directions.
This letter is the invitation.
He gave two million dollars to the place that hurt him. Not to punish it. Not to forgive it. To build something there — a dormitory, a room where the next displaced child sleeps in a bed instead of on a floor. The instinct is structural. You do not walk away from the wound. You build at the wound. You put a building on the exact piece of ground where the ground failed you.
Oregon State got fifty million. Stanford got thirty million. Oneida Baptist got two million and a dormitory with his name on it. Every gift maps to a coordinate on the biography. Every building rises at a place that shaped him.
Tacoma is the missing coordinate. The city where a nine-year-old boy arrived in 1973, unable to speak the language of the country he had just entered, and was sent away before the week was out. He has not come back. On the Portland Avenue corridor, in the Opportunity Zone that bears the data of exactly the kind of displacement he survived, a building is going up. It teaches people to use the technology he built. As of March 16, it proposes to run the agentic infrastructure he just released to the world. The circuit has been open for fifty-three years. It now closes in both directions — the emotional homecoming and the technical one. The software runs inside the building. The building sits in the city where the boy landed. The invitation is the same one it has always been.