265SmithWatt 75Neumann JHuangDHassabisFLiEMusk 20 Agentic AIforU
KingCharlesLLM DeepLearning009 NormanMacrae.net EconomistDiary.com Abedmooc.com
Update Taiwan AI june 2026- Nvidia 150 company ecosystem in Taiwan- 1000 agent CUDA-X platforms - ;;V10+8* (Vera Rubin 10) replaces G10 in strategic impact as Intelligence Nation Builders -example Japan role in V10; RTX spark transforms PC market; plan B if education systems continue to fail Jensen's Test, Cosmos Robot platform celebrating AI as greatest job creating leap ..jensen computex +14 pre-game scenesetters. Mooc for parents Celebrating Jensen Test
Huang Observing AI's biggest deals - 2012 1984 2026 & moments 1956 1905
https://www.youtube.com/watch?v=b1EaoN2CnCE Join Jensen live from Taiwan May 31 East Coast USA -pregame 9pm , jensen 11pm .. and meet Robots
June 1 in Taiwan -- pretrain with JENSEN's TEST
PREVIOUSLY
May 7 20000 DC brains thank SCSP AI+Expo. May 13 Trump asks Huang to join rest of China entourage at Alaska
Jensen video with stanford https://www.youtube.com/watch?v=tsQB0n0YV3k selected quotes
ED: Breaking 12 May 2026 AI Best kept secret- its all too much talk unless layer 5 AI connects massive application community scaling
GEMINI review 5/17/2026:: Governments and national champions in these regions are explicitly aligning their "Sovereign AI" policies with this layered approach to ensure they own their intelligence production.
Asia-Pacific Taiwan: Positioning itself as an "AI Island." It leverages its chip dominance (Layer 2) while investing over $3 billion in energy and "AI factories" (Layers 1 & 3). Singapore: Launched the first Model AI Governance Framework for Agentic AI in 2026 and is building a "Southeast Asia AI Plan" to localize the stack.Japan & South Korea: Rapidly building domestic "AI factories" to reduce dependence on foreign cloud providers.India: Implementing a Sovereign AI Plan that includes state-level LLM development and massive infrastructure buildouts. Indonesia & Vietnam: Emerging hubs for data center (Layer 3) and energy (Layer 1) partnerships to support regional AI growth
|
Middle East United Arab Emirates = Early adopters of the "Sovereign AI" narrative. The UAE’s AI Minister and NVIDIA have collaborated to frame AI as a "national natural resource". Saudi has made big purchases of Vera Rubin directly and with AWS
Europe France: Home to Mistral, which partnered with NVIDIA for a 18,000-chip data center to create a European "homegrown alternative". Germany & Netherlands: Focus on high-end manufacturing (ASML) and industrial AI applications (Layer 5). Nordics: Using their green energy surplus (Layer 1) to attract the world's most sustainable "AI Factories". UK: Prioritizing "compute sovereignty" through large-scale government-funded supercomputing projects
Gemini map review 5/17/2026continued here
May 5-9 Washington DC leaps closer to Mapping Huang AI. Even By Jensen Huang norms its been a massive weekend for democratising AI. Looks like 20000 turned up to Carnegie Melon commencement address Sunday May 10 , by Saturday 20000 DC brains had been greeted by layer 5 AI community apps: AI+expo www.scsp.ai
AI REVIEWS 10 MOST IMPOERANT AI (layer 5) APPLIUCATIONS FOR STUDENTS & TEACHERS OF 2025-6
Do any of top 10 surprise you? Some Westerners are surprised to see Robotics and Embodied AI at number 3. WE expect Asian supercities will lead with introduction not juts of self driving cars (almost ready after 15 years of AI research) but eg with public services in Tokyo where the elderly population dynamics cant find enough people to maintain the city world class usability standards - try benchmarking Bloomberg top 10 ai cities; The two academics with the deepest data algorithm fei-fei li and yann lecun have moved on to world ai and embodied ai. There's not much point in AI having the greatest brainpower in the world if a 2 year toddler understands gravity's consequences more actionably. Turns out its not just kids who need to celebrate Head, Shoulders, Knees & Toes (and arms); each requiring practice....
But where did vision that ai needs youth to scale its communal applications begin? As far as we can see one bleak December day CSIS Dec 2025 - please help us make sure joyous Diary of layer 5 AI is complete https://normanmacrae.ning.com/forum/topics/jensen100 Hopefully next school year AI curricula wont exist without it : JH CSIS December; India Ambassador DC CSIS Jan ; French & India Ambassador Brookings; India AI Summit Delhi; 2000 AI Expo www.scsp.ai 20000+ carnegie melon commencement
===================================
Breaking - help us video tour every placesmost youth inspiring intelligences
Breaking 16 March2026 at nvidia youtube today replay Jensen GTU speech - his global update of partners and purposes of computers with billion times more maths brainpower than separate human minds- note search out last time jensen keynoted in place you care about for more personalised ai agency nb jensen cases cover all tech needs to bring together to advance human work - eg 5 layers of full stack ai - energy , supercomputer chips, place unique investment purpose of tech and peoples, ai models, scaling community actions/aps. There are at least 3 gameboards beyond LLM agency: Einsteins' solutions to natures biggest puzzles such as illnesses and resource sustainability, space, how world class citizens demand last mile services from autonomous robots and eg drones.
AGENCY & OPEN CLAW GET 2026 OFF TO IMAGINATION GREATEST YEAR?
Harrison Chase | Co-Founder and CEO | LangChain
Jensen Huang | Founder and CEO | NVIDIA
Michael Truell | CEO and Co-Founder | Cursor
Misha Laskin | Co-founder and CEO | Reflection AI
Aravind Srinivas | CEO & Co-founder | Perplexity
Mira Murati | Founder and CEO | Thinking Machines Lab
Arthur Mensch | Co-Founder and CEO | Mistral
Daniel Nadler | CEO | OpenEvidence
Hanna Hajishirzi | Professor, Sr. Director NLP | AI2
Anjney Midha | Founder | Founder, AMP PBC
Robin Rombach | Co-Founder and CEO | Black Forest Labs
IN CHINA ITS THE YEAR OF THE LOBSTER (launching 300 AI team)
US 6G he has arranged for Nvidia to help US regain 6g leadership
US Last mile Transportation Autonomy With uber he's arranged for US to keep track of where world cllass cities are benchmarking both self driving cars and any last mile advances in autonomous machines
NB Nvidia Jensen is unique position to help with purposes because he has not only built the supercomputers AI performs on but cuda application libraries of most of the world leading uses of AI
|
Apps-Agents Million AI Mods Data Sov Invest Platforms Chips Energy ........ |
.. |
Views: 776
Continuation of Gemini review 5/17/26 of places where AI data sovereignty leaders understand 5 layer frame AI frame of Jensen Huang
2012 big deal (https://www.youtube.com/watch?v=tsQB0n0YV3k 1.05.34 May 2026 Stanford's best ever 2 minute case review - becoming worlds biggest corporation and humanity's 100 deepest partnerships) jensen huang " 2012 that's when two young engineers, Alex and Ilya, and Hinton, of course came up with a neural network model. And boom, it crushed the computer vision capabilities (imagenet annual world competition) of all the computer scientists, decades before them, in one shot. And so is that a big deal?
what else can you solve? And if this was able to solve some really amazing problems,
what does that mean to computers and computing? And so you just keep asking yourself these questions. And so you're just iterating like that, all the way to first principles. And then from that, you create a mental model about the future of computing.
|
How large would models become? And if so, what would computers look like?
Processing neural networks, how is that different than processing floating point numbers, and integers, and first principal mathematics?
We express everything in FP64 or FP32, but obviously, neural networks don't have to do that. And so you reason through it like this.
And then you build up a mental model of the future. And then your company, where you are going to be within it.
And then you just work backwards from there. And then now, the question, of course, is, you could be wrong.
|
1984 is when steve jobs launched personal macs to prevent orwell's 1984 and soon afterwards radiologist elliott fishman first began digital pattern analysis and mri became intelligence's greatest data space
ju,mp 20 years to 2004 nvidia jensen huang and steve jobs asked elliott if he'd like to join in coding pixels
( AI deepest (einstein pattern maths) consequences to date include crispr databases for dna editing - see berkeley's jennifer doudna or royal society paul nurse or Paris Emannuelle Carpentier; hassabis open sourcing 250 million proteins alphafold3
then jump to 2024 nvidia google and pixar now owned by disney startred partnering in robotics
=====================================
here is gemini's proofcheck of this case study
The Johns Hopkins & Elliot Fishman Connection
Jensen Huang and Dr. Elliot Fishman share a friendship and professional history that dates back decades, rooted in their mutual interest in 3D visualization.
Johns Hopkins Medicine +5
Why 2004 was the Turning Point
The reason your memory of 2004 is so sharp is because that is when NVIDIA began developing CUDA (Compute Unified Device Architecture).
Wikipedia
holisticon.pl +2
Key Milestones in the "NVIDIA + Radiology" Timeline
Era |
Focus |
Key Figure / Tech |
Late 80s |
Early 3D CT |
Elliot Fishman using Pixar hardware |
2004 |
The Pivot |
Development of CUDA begins; Gelato launched |
2007 |
Accessibility |
CUDA officially released, making GPU coding "easy" for scientists |
2010s |
Photorealism |
Launch of Cinematic Rendering for "lifelike" medical scans |
Today |
AI Agents |
NVIDIA MONAI framework used for AI-driven diagnostics |
💡 Jensen's Vision: Huang often notes that radiology was the first field many "experts" predicted AI would destroy, but he argues it has instead become the "superpower" that allows doctors like Fishman to see what was once invisible.
If you're interested, I can look into:
Dec 1, 2019 — The Elliot K. Fishman, M.D. Professorship in Radiology was established in the Russell H. Morgan Department of Radiology and Radiol...
Johns Hopkins Medicine
CUDA (Compute Unified Device Architecture) is a proprietary parallel computing platform and application programming interface (API...
Wikipedia
Elliot K. Fishman, M.D. is Director of Diagnostic Imaging and Body Computed Tomography (CT) in the Russell H. Morgan Department of...
CTisus
AT CROSSROADS/CROSS-SEAS TO HUMAN INTELLIGENCE
2026 Taiwanese people with 150+ suppliers to Nvidia Vera Rubin Supply Chain are the trusted heart of best AI apps and agency scaling community's most vital actions
SUPPLY CHAIN SERVING HALF TRILLION DOLLARS GREATEST COMPUTE BRAINS ...
https://www.youtube.com/watch?v=E-F5-0_-WGY Fox business intervew jensen hunag may 2026 Nvidia quarter Here's what jumped out. 51.2 billion of your 57 billion revenue number came from the data center chip sales. Monstrous: I'm so curious to know what's the largest single order to date that you've gotten.
Which of Nvidia's Eastern Data Sovereign partners want most help with UN health goals?
Currently deep ai advances are being celebrated drug discovery - see nvidia clara platforms (footnote 1) parallel to 250 million protein breakthroughs from hassabis deep mind. IN May 2026 Hassabis announced hope that Einstein Brain will be replicable by ai agnets within 3 years - leadin to a lot wider ai applications- here are 19 joy of ai contexts chats hope einstein brain will help them link humans to
.
For their own people Japan and Korea seem to want support of mental health and of course plague mitigations; japan may be leader in helping UN agenda in africa
Taiwan and hong kong appear to make quick start in uniting partnerships around digital twinning big cities most affordable teching hospitals
=======================================
some questions to update
emerging health ai atgtu 2022
March 2022
MORE INCEPTION/nVIDIA CLUB ai STARTUPS
Jensen Huang
AGENTIC AI TAIWAN ECOSYSTEM/Supply Chain _ Nvidia &
Pre-game hosts Bruce Lu (Goldman Sachs) and Tracy Tsai (Gartner)
The AI Server Boom YJ Mii, TSMC Barry Lam, Quanta Computer Simon Lin, Wistron
Journey From PCs to Embedded AI Haw Chen, Techman Robot Jonney Shih, Asus Rick Tsai, MediaTek
Agentic AI and Robotics Transforming Taiwan Industry Simon Chang, Delta Electronics T.H. Tung, Pegatron Kathy Yang, Foxconn
Building AI Infrastructure for Taiwan Hung-yi Lee, NTU Neo Yao, Visionbay.ai Alex Yeh, GMI Cloud
X grok says these are 30 clusters of platforms its most familiar with but agentic ai is a fast changing world
Foundational / Math & Compute Clusters (used everywhere)
cuBLAS / cuTENSOR / CUTLASS — Linear algebra & tensor ops (core for all ML/DL).
cuDNN — Deep neural network primitives (training/inference backbone).
TensorRT (incl. TensorRT-LLM) — High-performance inference optimizer/runtime.
Thrust / CUB / CCCL — Parallel algorithms (sort, scan, reduce, etc.).
cuSOLVER / cuSPARSE — Dense/sparse solvers & linear algebra.
cuFFT / cuRAND — FFTs and random number generation.
NCCL / NVSHMEM — Multi-GPU/multi-node communication.
Data Processing & ML Clusters
RAPIDS (cuDF, cuML, cuGraph, cuVS) — GPU pandas/scikit-learn/NetworkX equivalents (data science & graphs).
DALI — Data loading/preprocessing for vision/audio.
NeMo (Framework + Curator) — LLM training, customization, RAG, agentic AI.
youtube.com
cuOpt — Optimization/routing (logistics, planning for agents/robots).
Computer Vision & Media Clusters
NPP / CV-CUDA / cuCIM — Image/signal processing & biomedical imaging.
OpenCV (GPU-accelerated) + nvImageCodec — Classic CV + high-throughput codecs.
Video Codec SDK + Optical Flow SDK — Video encode/decode & motion.
nvJPEG / nvComp — Compression & image handling.
Autonomous Driving / Self-Driving Cars Clusters (NVIDIA DRIVE ecosystem)
DRIVE OS + DRIVE AGX platforms — Full AV stack (perception, planning, control).
Cosmos (World Foundation Models) — Physical AI simulation & prediction for AVs.
developer.nvidia.com
cuLitho (lithography) + sensor fusion tools — Supporting manufacturing & real-time perception.
Vision-Language-Action (VLA) models (e.g., Alpamayo series) — For end-to-end AV autonomy.
siliconangle.com
Robotics & Physical AI Clusters (NVIDIA Isaac ecosystem)
Isaac Sim + Isaac ROS — Simulation, ROS 2 integration, manipulation.
Isaac GR00T / Humanoid models — Foundation models for robots.
siliconangle.com
Warp / PhysicsNeMo — GPU physics simulation & AI-driven physics.
cuRobo — CUDA-accelerated motion planning/kinematics/collision.
github.com
Manipulator / AMR workflows — Libraries for arms, mobile robots, perception/SLAM.
Agentic AI & Enterprise Clusters
NVIDIA AI Enterprise / Nemo Retriever — Agent blueprints, RAG, semantic search.
youtube.com
Morpheus — Cybersecurity & real-time streaming AI pipelines.
Riva — Speech/translation AI (multimodal agents).
Earth-2 / cuQuantum — Climate/physics simulation + quantum (advanced modeling for agents/robots).
Advanced / Specialized Clusters
CUDA-Q (Quantum) — Hybrid quantum-classical for future optimization/simulation.
Partner/Integration stacks (e.g., PyTorch/TensorFlow + CUDA, Triton Inference Server, GPUDirect Storage) — For scalable deployment of agents/robots.
Developers pick combinations based on the sector: e.g., AV stack = DRIVE + Cosmos + TensorRT + DALI + cuRobo-like tools; Robotics = Isaac + Warp + cuDNN + ROS 2; Agents = NeMo + cuVS + TensorRT LLM + cuOpt.
===============================
AND HERE IN TAIPEI IS WHERE IT ALL BEGINS 3.40
[JENSEN HUANG] Welcome to GTC, Taiwan. So great to see all of you. Very good to be home. I brought my parents home. Where are my
parents? Everybody give a round of applause to my mom and dad. And a round of applause
for our pre-game show superstars, ladies and gentlemen. Look how adorable they are.
The superstars of Taiwan. There are so many of you here today. We
are broadcasting this right now to 70 other launch parties across Taiwan. 70 different conferences
are going at the same time. Everybody is watching this live. We have so much to tell you, and I
have so many partners to thank. It is incredible how large our ecosystem in Taiwan has become.
Most of the time, when people think about ecosystem, they think about our software stack.
They think about the developer ecosystem above the computing systems that NVIDIA builds. But
NVIDIA’s ecosystem spans all the way upstream to all of our supply chain here in Taiwan,
where it all begins, and downstream all the way to data centers, and eventually to end users.
Today we’re going to talk about almost all of the ecosystem. There are so many people to thank.
I love my ecosystem here. So many incredible companies, some of my favorite ecosystem partners.
Taiwan’s rich ecosystem — the world’s best supply chain ecosystem. Unbelievable.
Well, thank you all for being here. And this year, our businesses together are
growing incredibly. In fact, somebody told me last night that the annual GDP of Taiwan is
going to grow almost 10%. Unbelievable. Well, we have a lot to talk about. Let’s get going.
Two years ago, when I was here, I started to talk to you about how AI has moved from generative AI,
and other waves of AI that are coming. The next wave of AI was agentic AI. And
today we can say that agentic AI has arrived, that useful AI has arrived.
Now, what does this mean? This is GitHub. This is, of course, one of the first applications of agentic AI — software coding.
One of the most valuable professions, an incredibly large ecosystem — 30 million,
40 million professional software developers, probably another couple hundred million who are
students and enthusiasts and so on and so forth. But say 30, 40 million software developers in the
world code for a living. And this represents most of them. This is GitHub. The pull request
is when they download software, they modify it, and commit is when they push it back up. And so,
if you could look at this — in 2023 the number of commits was 300 million, in 2024,
400 million, in 2025, 500 million commits. In the first few months of 2026, it has nearly tripled.
Now what does that mean? 30 million software developers representing about $3 trillion worth
of GDP — that is what they’re paid, $3 trillion worth of salaries per year, which is generating
economic growth for the rest of the industries. Say $100 trillion of the world’s industries is
impacted — generated by $3 trillion worth of salary. That $3 trillion worth of salary is
now producing nearly three times as much output. It’s effectively $9 trillion of productivity from
$3 trillion of salaries. Does that make any sense? The difference is absolutely extraordinary. This
is the potential. This is the promise of AI.
The number of software engineers is actually
increasing. People talk about AI reducing jobs — complete nonsense. It’s causing more
software engineers to be hired, and the reason for that is very simple. If you can hire a software engineer and you can generate $9 trillion worth of productive work,
why wouldn’t you want to hire more software engineers? If that line was flat, then obviously
people will hire fewer software engineers. But because the output is so incredible, people want to hire more software engineers. This is going to show up in our economy very soon.
And so, the first thing is: useful AI has arrived. Now, what does that mean from the industry’s
perspective? That means that tokens are now in extraordinary demand. Because if you can do this,
you’re going to want to produce more of it. And because tokens are now profitable units — tokens
are now profitable units of revenue. Because it is now profitable, the AI companies want to build a
lot more tokens, generate a lot more tokens, build more AI factories, which is the reason why compute demand here in Taiwan has skyrocketed. It is precisely the reason why all of you are so busy, and your businesses are doing so well. In fact, that looks like some of your stock prices.
The compute pattern has changed. Everything has changed. So the first idea is that useful
AI has arrived. AI is now a profit generator. AI is now a GDP generator. The whole
new kind of computing pattern — not just a large language model, but an agent. Today, almost everything we’re going to talk about is going to be based on this.
So let me take a quick moment and show you what I’m talking about.
This is an agent — an agent. application. In the old days, this would be an application — this would be code,
11 minutes, 15 seconds
and this would be an operating system. Application, code running inside an application,
inside an operating system. Today, it is an agent, which consists of a large language model, or many, sitting inside a harness, and that harness helps orchestrate it to do productive work.
This is the input — when that input comes, it has to understand, observe, reason, act,
use tools. That tool could be a spreadsheet, a web browser, a data processing engine, a database
engine, for example. This is orchestrated — this harness orchestrates this routing of information.
Every single time it touches — either processing the context, understanding what is happening,
reasoning about what to do, coming up with a plan that it acts on — that orchestration is managed by software. And so this is fundamentally an agent. It deals with short-term memory — called working memory — and long-term memory, just like we do. And so the memory management system is incredibly important. This entire system is called an agent. The large language model is used to do the thinking, and the harness connects everything together, just like an operating system.
And so this is the new computing model. This is what an agent can do — incredible
things. This is the big breakthrough: the simultaneous convergence of large
language models that are now able to do a really good job thinking, reasoning, planning, using tools, and the fact that we now have these harnesses that manage memory,
manage orchestration, use tools. We can now do amazing things. Let me give you some examples.
This is a prompt. This is the code that is generated, and this comes out. You guys — what
do you think? Pretty amazing, right? Look, we use Claude Code here, but Codex does an incredible
job as well. Here’s another example. This is the input: “Create a GIF — NVIDIA green dots on black,
scatter, form Taiwan 101 building, NVLink GTC Taipei 2026, morph to NVIDIA logo, then scatter,
repeat.” Right? So you saw that — that was the prompt. Here is the next one: “I lost my remote
control battery clip. It looks like this. Create a CAD file.” It uses a tool, creates a CAD file,
ready for 3D printing to create a new one. Makes sense? This is now the new computing pattern.
Whereas we used to launch an application, click and type, we now replace that with explaining
to the AI what we want — our intent — and the AI generates the code, or uses tools,
and produces the necessary output. This is how computers are going to work in the future.
This is agentic AI. For two years we’ve been building towards this, and now it has arrived.
Now, one of the big breakthroughs of course is tool use. A lot of people have said, “Jensen,
agentic AI is coming, therefore all of the software companies are going to go out of business.” I said: it’s exactly the opposite. Because there are going to be so many agents,
the world is no longer limited by the number of people. Therefore, those agents are going
to use more tools than ever. This is actually an incredible time to be a software company.
But the software has to be presented to the agent in a way that the agent can use it.
This is a big breakthrough, and in fact, what we have done — what NVIDIA’s treasure is — is
all of our CUDA libraries. I call them CUDA-X libraries. This is NVIDIA’s treasure. Today,
we are able to now present these CUDA-X libraries to agents who can use them much more effectively than even humans. And so, this is a wonderful time for CUDA-X libraries. Just take a look. 20 years ago we built CUDA, a single architecture for accelerated computing. We reinvented computing.
A thousand CUDA-X libraries help developers make breakthroughs in every field of science
and engineering. CUDA-X libraries are tools for agents. cuLitho for computational lithography.
cuOpt for decision optimization. cuDSS for direct sparse solvers. AI-Q for deep research across
structured and unstructured documents. Aerial for AI RAN. PhysicsNeMo for differentiable physics.
Parabricks for genomics. At their foundation are algorithms, and they are beautiful.
A round of applause for math — math is beautiful. The computing pattern of software is going to
change. In fact, let’s come back to this. This is the agent — the ultimate disaggregated and
distributed computing model. So many different computers are going to be activated in order to
process this agent. The agent consists of model, harness, tools, and skills, and a runtime. All
of that is running at different places in a data center. You can think of the model as the brain,
the harness as the body, and the tools that it uses, working in a runtime — think of it
as a workshop. So this is a person, a worker, working with tools in a workshop. Of course,
this is being done at extraordinarily large scales. And each one of those steps is running
in a different part of the computer. You can see the large language model is
thinking — context processing, observing, understanding the environment, reasoning,
coming up with a plan, and acting on the plan. Every single time that happens, an entire rack
of Grace Blackwell NVLink 72 is activated. It’s thinking with a large language model.
Whenever it uses a tool, a CPU is used. That tool could be a C compiler. It could be Python. It
could be JavaScript. Or it could be accelerated computing. Today’s agents are relatively simple
users of tools. Tomorrow, they’re going to be very sophisticated users of tools, which is the reason why the CUDA-X libraries that I showed you are going to be incredibly
popular with agents. They solve some of the most important problems the world knows. And all of
our CUDA-X libraries are now going to come with skills that the AI could learn how to use. So the
CUDA-X library comes with skills — basically a manual — the AI reads it and goes, “Aha,
that’s how you use it.”
The ability to use these libraries by agents is going to be incredible.
And so, the tools run on CPUs and GPUs and large language models. The security harness
runs on CPUs and a security processor called the DPU — NVIDIA’s BlueField. The orchestration of all
this runs on a CPU. This is the entire harness — the CPUs orchestrating all of the work. One of the
hardest parts is memory. You can just imagine: the working memory — called KV caching — is about what to remember. Compassion, not just compression, but how to retrieve. Do you retrieve structured data? Do you retrieve unstructured data? What is the ontology, the relationship of all of these different data structures to itself? That entire processing is incredibly complicated. The memory system of AI is going to cause the storage system to be completely revolutionized.
As you can see, every aspect of this computing model — this computing pattern,
this new application called an agent — is fundamentally different than the way that applications used to run: a whole bunch of software sitting inside a binary, sitting
inside an operating system. This is the reason — this disaggregated, this distributed, this
heterogeneous computing problem — is precisely the reason we built our next generation: Vera Rubin.
Vera Rubin is not one chip. Vera Rubin is not a GPU only. It starts with a GPU,
But Vera Rubin is incredible. This entire thing is Vera Rubin. From end to end, it has
GPUs — Vera Rubin NVL72. It is orchestrated by Vera CPUs that I’m going to tell you more about.
The storage systems — revolutionary. Vera, along with ConnectX-9, our software stack called DOCA, the security processor that’s inside so that everything is encrypted at rest,
in motion, as well as in use. Everything across this is secure because the AI model
is so precious. This is the reason why this entire system observes confidential computing.
Each one of these systems would be a complete revolution in itself. Vera Rubin is the most
ambitious endeavor in the history of our company. The whole company worked on Vera Rubin — across all 40,000 engineers, not to mention all of you. All of you participated in the creation of this
entire system.
Vera Rubin is really a miracle, and it’s not just one chip — it is so many.
Well, it’s even beyond that. A long time ago, NVIDIA used to be a GPU company. But over the
years we’ve evolved to become a systems company. You’re looking here now at the most complex
system, the most complex ground-up system ever designed. But ultimately, our customers and our
partners don’t want to buy computers. They want to build AI factories, which is the reason why
NVIDIA has really started to transform ourselves yet again. You can see so much of our technology
is now at the entire infrastructure scale. Our partners are at infrastructure scale — power
generators, cooling systems, the grid providers. So many industrial companies are now part of our
ecosystem, because ultimately we’re trying to build an entire stack — just like GPUs, just
like when we were building Grace Blackwell NVLink 72, just like now, we are building a full-stack
system so that our customers could build amazing AI infrastructure. Let’s take a look. ________________ [VIDEO NARRATION]
The world is racing to build AI factories — the largest infrastructure buildout in
human history. AI factories are incredibly complex. Every layer — chip, rack, network,
power, cooling, grid — must be designed together from end to end, because compute is revenue.
NVIDIA DSX is the blueprint: a reference design for building and operating AI factories at
maximum efficiency and profitability. It starts with DSX Sim — the Omniverse
blueprint. Partners design and validate an NVIDIA Vera Rubin AI factory before a single
rack is ordered. They plan the layout, simulate the power and cooling, design the network,
validate every integration, test every change in the digital twin. The factory powers on.
DSX OS takes over — it provisions, operates, monitors, and remediates the infrastructure,
turning the installed systems into trusted, multi-tenant, resilient, AI-ready capacity.
Today’s AI factories overprovision power by up to 40%. DSX MaxLPS lets operators
safely deploy more GPUs inside the same power budget, adding billions in annual revenue.
Breakthrough hot liquid cooling at 45 degrees Celsius uses less water and energy — more power going to revenue-generating compute. Incredible.
Dynamic power allocation steers power from rack to rack, recovering stranded watts,
sending them where work is happening. In-rack power smoothing flattens peak current spikes
and power surges. Throughout the factory, teams of AI agents work with DSX MaxLPS,
continuously coordinating to balance cooling and power to meet workload demand.
DSX AI factories are flexible energy assets that operate cooperatively with the grid. DSX
Flex reads real-time grid signals and dynamically adjusts back to power when the grid needs relief.
100 gigawatts of AI factories will come online before the end of the decade. NVIDIA DSX AI
factories run at highest efficiency, produce the lowest-cost tokens, and make the grid stronger.
around 29.48
________________ [JENSEN]
I’ve shown you ecosystem slides of the past, where NVIDIA’s computing layers and software
stacks are integrated into other people’s platforms — third-party platforms and libraries
that serve end markets. That was a computing ecosystem. This is an AI factory ecosystem.
This is way downstream of all of you. Upstream of me is all of you, and downstream of us is this
ecosystem. Because NVIDIA, ultimately, is not just building a GPU, not just building a system — we’re
helping customers build these AI factories, this AI infrastructure, that is so immensely complex.
Each one of these, at one gigawatt level, started at $20, $30 billion. It is at $50,
$60 billion, and soon it will be $80, $100 billion per gigawatt. $100 billion into an
AI factory. It must work the first time, and it must work right away. The cost of capital
is incredible. The complexity is incredible.
So as you see, we used to design a chip inside
a computer. And then we simulated a system inside a computer. Today, you saw just now — everything was built in Omniverse. I’ve been working with Omniverse with all of you for a long time.
This was the dream come true, so that we can build these gigantic systems — as large as the
world wants to build — inside a digital framework, inside a digital simulator,
in a digital world, long before we break ground and put our money to work.
So this is our ecosystem. We call it DSX. RTX is for our GPU, DGX is for our systems,
31:42
31 minutes, 42 seconds
and now DSX — basically infrastructure. Because of the work that we do here,
across this entire stack including our systems and software, it’s the reason why we can work
with small companies and enable them to be world-class AI clouds.
Every one of these
I’m about to show you are small companies — just recently.
see next post
Above jensen said
DSX — basically infrastructure. Because of the work that we do here,
across this entire stack including our systems and software, it’s the reason why we can work
with small companies and enable them to be world-class AI clouds.
-here are some cases of startups leaping with nvidia's agency platforms or partnering agency supply chain
And now CoreWeave is worth $50, $60, $70 billion, and growing incredibly fast.
Recently we worked with Nebius — and again, they’re growing incredibly fast. Each one of these clouds has incredible customers:
Cursor, the software coding company; Black Forest Labs, image generation; World Labs, world foundation models; Revolut, the leading financial services AI company; and Shopify. Here’s another one — Nscale, and their customers are British Telecom, Google — Google is using one of our AI clouds.
Korean Thinking Machines, a frontier labs company, very exciting. Here’s NAVER Cloud in Korea — Bank of Korea, Hyundai, so many incredible companies.
Here’s one in India — Yotta. Incredible companies.
Here’s one based in Singapore, building in Australia — Together AI, AI Singapore. And here’s one
in Indonesia Indosat
Each one of these companies are serving regional as well as global customers.
AI is going to run everywhere. Every company will be powered by it, every region will
here in Taiwan, GMI.
Incredible, incredible companies, incredible opportunity, but all of them need several
things. Of course, they need the computing stack — this entire stack underneath. This is what made
NVIDIA famous. All of our hardware and software and libraries, our connection into the world’s
ecosystem of third-party developers, makes it possible for anyone to stand up an AI cloud.
However, the AI cloud is so complex now. This is the software version, the computer science
version. The money version, the asset version, is what I showed you earlier — it’s a giant factory.
Having this ability alone is not enough, which is the reason why NVIDIA has become
an AI infrastructure company. Doing this well — becoming incredibly good at helping customers
build AI factories and deploying AI factories — is incredibly important, and the reason for that is
this: compute is revenue now. Compute is profit. The absence of revenue and profit is loss.
And so it’s really important to realize — this is an example of AI infrastructure coming online. It
could be coming online quickly, it could take a while. Its throughput could be high, it could
be low. Its resilience, or reliability, could be good or bad. And its lifetime of usefulness could
be long or short. Because this represents $50, $60 — going to $100 billion
— this curve matters
greatly, which is the reason why NVIDIA is such a great partner. Working with us, because of our
fully integrated capability — we didn’t just come up with a PowerPoint slide. We created the entire
infrastructure. We connected everything together. We built out billions and billions of it ourselves
to make sure that everything works well. As a result, our time to first token,
our time to first inference, our time to training turned on is much faster. Second,
because our throughput per watt — tokens per watt — is utterly world class. And the reason
for that is because we integrate everything, we design everything from the ground up, we simulate the entire system, and we use extreme co-design. Just like I showed you just now with
the Vera Rubin rack — everything was designed in order to deliver on this incredible throughput.
If your data center, if your factory, has one gigawatt, it will not have more — that’s all
the power generation you could have. If you have one gigawatt of power, then throughput per watt is revenue. Because every token is profitable. Every token is revenue. This is the future. Compute is
revenue. Performance per watt is your revenue. Choosing the wrong architecture just because
the chips are cheaper doesn’t translate. Doesn’t make sense. You need to make sure of your revenue per watt. The more you buy, the more you make. And then third is reliability. If you ever get a chance to see these data centers, there are so many moving parts, millions of cables. The ability for all of those computers to work harmoniously, reliably,
is extremely hard. We have now been operating at very large scale for a very long time. That
experience matters. That difference — mean time between interrupts — is extremely important.
And then lastly, this is very hard: the lifetime of these systems. The software is
changing all the time. Four years ago — the time of Hopper — AI had completely changed.
Six years ago — the timeframe of Ampere — AI had completely changed. We started out talking
about CNNs. Then we talked about Transformers. Then we talked about mixture of experts. Now
we’re talking about agentic systems. Every single generation, every single few months, the software
industry is coming up with new technology. If your architecture is not flexible, if your ecosystem is
not rich, then this curve cannot be long. You cannot predict how long your system can last.
39 minutes, 6 seconds
I can. NVIDIA systems are all over the world. Software developers start with NVIDIA CUDA.
39:14
39 minutes, 14 seconds
And by definition, therefore, the life — the ecosystem, the useful asset — is going to be
39:21
39 minutes, 21 seconds
much longer. The difference is essentially cost. You could think of it as revenue, but the other
39:28
39 minutes, 28 seconds
side of revenue is cost. If the life of the asset is long, the TCO is low. This is the difference.
39:37
39 minutes, 37 seconds
This is what it looks like when compute works this way —
39:44
39 minutes, 44 seconds
the more you buy, the more you make. Now, all of you are experiencing this with me, isn’t
39:59
39 minutes, 59 seconds
that right? All of your demand, your factories are working so hard, your people are working so
40:06
40 minutes, 6 seconds
hard all across Taiwan, because everybody wants to make money. They realize that useful AI is
40:16
40 minutes, 16 seconds
here. Profitable AI is here. Compute demand is incredibly high, and compute demand is the
40:26
40 minutes, 26 seconds
constraint. And so, let’s all work super, super hard and help the world stand up AI factories
40:32
40 minutes, 32 seconds
everywhere. This is why it’s so important. I am so happy. Here I am, standing in front of
you — Vera Rubin is in full production. Vera Rubin is in full production.
40 minutes, 52 seconds
The supply chain we created for Vera Rubin is twice as large as Grace Blackwell. It’s
incredible. And what used to take two hours to assemble one Grace Blackwell rack now only takes
five minutes. So not only is the capacity higher, the throughput is a lot faster,
and we need it all to support the demand. This ecosystem is extraordinary. Millions
of square feet has been put online to support Grace Blackwell and preparing now — ramping
up now — Vera Rubin. I want to thank all of you. Vera Rubin is now in full production. Thank you.
Insert you would need to verify with nvidia bot Gemini sees 26 core supply chain system members of Vera rubin as of May 2026
this massive operation, NVIDIA leverages hundreds of ecosystem partners worldwide—including 150 in Taiwan alone—across 350 factories. [1, 2, 3]
The 26 core official supply chain partners showcased as the primary builders, component makers, and infrastructure providers for the Vera Rubin architectural platform include:
Instead of single-sourcing, NVIDIA has officially qualified all three major memory titans for its advanced sixth-generation HBM4 slots: [1, 2]
These are the leading system integrators responsible for deploying standard configurations and full scale [NVIDIA MGX rack architectures](https://investor.nvidia.com/news/press-release-details/2026/NVIDIA-... Into-Full-Production-to-Power-Agentic-AI-Factories-Worldwide/default.aspx "NVIDIA MGX Architecture Partner Support"):
Would you like to break down how the Taiwanese sub-tier suppliers are specifically handling the new liquid-cooling requirements for these racks, or look closer at the Vera CPU's architecture?
Continuing Jensen-41 minutes, 41 seconds
Let’s take a look. ________________ [VIDEO NARRATION] Large language models
41:47
41 minutes, 47 seconds
generate answers. Now, AI agents can do work. But processing agentic AI is a whole different kind
41:55
41 minutes, 55 seconds
of problem. Agents observe, reason, plan, use tools. They manage massive context,
42:02
42 minutes, 2 seconds
juggling working memory and long-term memory. They spin up subagents — specialists on demand.
42:08
42 minutes, 8 seconds
NVIDIA Vera Rubin is a multi-rack, pod-scale system built to process agentic AI,
42:14
42 minutes, 14 seconds
and is now in full production. The manufacturing, automation, and orchestration across the supply chain — a miracle to witness. Our journey started when
42:24
42 minutes, 24 seconds
we launched the first AI supercomputer, NVIDIA DGX-1. Over the next decade, we pushed every
42:31
42 minutes, 31 seconds
chip and system to the limit — from Pascal and the first Ampere, to Grace Blackwell,
42:36
42 minutes, 36 seconds
the first rack-scale AI supercomputer. And now, Vera Rubin — the first multi-rack, pod-scale
42:43
42 minutes, 43 seconds
supercomputer built for the agentic age. It starts at TSMC. The seven new chips
42:49
42 minutes, 49 seconds
that make up Vera Rubin take shape through hundreds of processing steps.
42:54
42 minutes, 54 seconds
Three-nanometer process. CoWoS advanced packaging. HBM4 memory from Micron, SK Hynix, and Samsung.
43:03
43 minutes, 3 seconds
The Vera Rubin GPU — six trillion transistors with over 18,000 components on one board.
43:10
43 minutes, 10 seconds
Vera Rubin NVL72 does the thinking — prompt and context understanding, reasoning, and planning.
43:18
43 minutes, 18 seconds
Next, a new modular compute tray — streamlined with a new PCB midplane. ConnectX-9,
43:26
43 minutes, 26 seconds
SuperNICs, and BlueField-4 DPUs — all maintenance-accessible with no cables,
43:32
43 minutes, 32 seconds
for resiliency and AI factory scaling. 18 compute trays, nine hot-swappable NVLink switch trays. New
43:40
43 minutes, 40 seconds
high-efficiency liquid-cooled bus bars carrying over 5,000 amps — the equivalent of 20 electric
43:47
43 minutes, 47 seconds
cars at full acceleration. Together, 1.3 million components form this third-generation MGX rack.
43:56
43 minutes, 56 seconds
Congratulations to Microsoft for their operational Vera Rubin NVL72 engineering rack. Congratulations
44:02
44 minutes, 2 seconds
to Dell and CoreWeave as well for standing up their Vera Rubin NVL72 engineering racks.
44:08
44 minutes, 8 seconds
Then, the Vera CPU rack — 256 CPUs in a single liquid-cooled rack, orchestrating the models,
44:17
44 minutes, 17 seconds
shuffling memory, launching tools. Foxconn and Quanta brought the Vera LPX rack to
44:24
44 minutes, 24 seconds
shape — 256 Groq LPUs across 16 trays, 40 petabytes per second of SRAM bandwidth — for
44:32
44 minutes, 32 seconds
ultra-low latency. While NVL72 generates tokens at the highest throughput, the
44:39
44 minutes, 39 seconds
LPX rack generates them at the lowest latency. Vera BlueField-4 STX — where AI keeps its memory.
44:47
44 minutes, 47 seconds
Storage processing accelerated by BlueField-4, connecting memory,
44:52
44 minutes, 52 seconds
storage, and in-silicon security. And NVIDIA Spectrum-X Ethernet Photonics — the
44:59
44 minutes, 59 seconds
world’s first Ethernet switch with 200-gigabit co-packaged optics. TSMC’s CoWoS process,
45:06
45 minutes, 6 seconds
chip-scale packaging, and ultra-high-powered laser dies on indium phosphide.
45:13
45 minutes, 13 seconds
Vera Rubin — five connected rack-scale systems, a supercomputer for AI agents.
45:20
45 minutes, 20 seconds
150 supply chain partners across Taiwan. Millions of square feet of factory floor,
45:25
45 minutes, 25 seconds
hundreds of sites. Chips, packages, systems, and data centers pushed to the limits of size,
45:33
45 minutes, 33 seconds
power, and scale. This is what we call extreme co-design. We did this with Taiwan. Together,
45:39
45 minutes, 39 seconds
we reinvented computing for the age of AI. Taiwan was with us at the beginning,
45:44
45 minutes, 44 seconds
and here today, as we bring Vera Rubin to the world. Thank you, Taiwan.
45:51
45 minutes, 51 seconds
45:51
45 minutes, 51 seconds
________________ [JENSEN]
45:57
45 minutes, 57 seconds
Ladies and gentlemen, Vera Rubin. Vera Rubin was not built just to run
46:09
46 minutes, 9 seconds
46:19
46 minutes, 19 seconds
is the reason why the agent architecture is the last great computer science breakthrough. It has
46:25
46 minutes, 25 seconds
taken this many years for agents to realize their potential and become useful. It stands
46:30
46 minutes, 30 seconds
to reason that the computer that runs it is the most advanced in the world. This is Vera Rubin.
46:36
46 minutes, 36 seconds
Let’s take a look. Can we bring out Vera Rubin, please?
46:59
46 minutes, 59 seconds
And Janine, do we have the racks, the systems? It looks heavy. This is Vera Rubin — Vera Rubin
47:14
47 minutes, 14 seconds
NVL72. This is the LPX rack. At the next GTC, I’m going to talk to you about a lot more of
47:22
47 minutes, 22 seconds
this. Today, we have so much to talk to you about. This is the Vera CPU rack — 256 CPUs,
47:30
47 minutes, 30 seconds
all liquid-cooled. Let me tell you about Vera in just a moment. This is the Vera BlueField
47:37
47 minutes, 37 seconds
storage processing system — and also security system. And of course, this is our Mellanox
47:44
47 minutes, 44 seconds
networking — the world’s first CPO. This is Vera Rubin. Incredible technology all coming together.
47:53
47 minutes, 53 seconds
Now, when we built Hopper, we built Hopper — as you know — for pre-training. Pre-training
48:00
48 minutes
was the most important application, the most important workload we were working on at the time. Then when we worked on Grace Blackwell, everybody said, “Jensen, you know,
48:10
48 minutes, 10 seconds
NVIDIA is really good at pre-training. Inference is so easy.” Do you remember that? People used
48:18
48 minutes, 18 seconds
to say, “Inference is so easy. We could do that too.” But as you know, inference equals money,
48:24
48 minutes, 24 seconds
and the models — MoEs — are so complicated, and to do it at incredibly high response time,
48:33
48 minutes, 33 seconds
fast interactivity, and high throughput at the same time is incredibly hard.
48:38
48 minutes, 38 seconds
Which is the reason why we created NVL72. Today, NVIDIA’s token cost is the lowest in
48:45
48 minutes, 45 seconds
the world — not by 10%, but by multiples, orders of magnitude. All because we did
48:53
48 minutes, 53 seconds
extreme co-design. All because we understood the computing model, the computing pattern
48:59
48 minutes, 59 seconds
of inference, and we were able to create NVL72. Now with Vera Rubin, it is beyond inference — it
49:09
49 minutes, 9 seconds
is now inference in an agentic system. This is Vera Rubin. No cables, no hoses, no fans. What
49:22
49 minutes, 22 seconds
used to take — the last time I showed this to you — cables everywhere, amazing to look at. But now
49:32
49 minutes, 32 seconds
there’s a PCB in the middle which connects both sides. What used to take two hours now takes five
49:39
49 minutes, 39 seconds
minutes. The reliability and the resilience of Vera Rubin is going to be off the charts.
49:46
49 minutes, 46 seconds
This is our Vera CPU tray — the most advanced CPU that has ever been built. I want to show
49:53
49 minutes, 53 seconds
you that in just a second. And this is our storage tray — two Vera CPUs, four ConnectX-9,
50:04
50 minutes, 4 seconds
incredible amounts of storage. This is our new LPX — the Groq LPX rack, designed for very low
50:15
50 minutes, 15 seconds
latency inference. The throughput is delivered by Vera Rubin and extended with NVL72. If you want
50:24
50 minutes, 24 seconds
to extend that even further, you can add Groq LPUs. Here we have the Vera Rubin NVLink switch
50:33
50 minutes, 33 seconds
tray — the switches in the middle — and this is revolutionary. Because of Vera Rubin, because
50:40
50 minutes, 40 seconds
of NVL72, and the NVLink switches that we created and invented. And this is our Ethernet switch for
50:50
50 minutes, 50 seconds
scale-out. What’s amazing is we introduced these two systems for Grace Blackwell, and today NVIDIA
51:03
51 minutes, 3 seconds
is the largest networking company in the world. I’m so proud of the networking team. This is such
51:11
51 minutes, 11 seconds
an incredible enabler for everything that we do. I want to now talk to you about the next major
51:17
51 minutes, 17 seconds
industry we’re going to be part of. Thank you, Janine. It’s right here —
51:35
51 minutes, 35 seconds
I think there are 2,000 people back there pulling them.
51:44
51 minutes, 44 seconds
Okay, let’s talk about CPUs. Vera CPUs. CPUs built for the age of agentic AI.
51:54
51 minutes, 54 seconds
All of the CPUs until now were created for people. We were the users, we were the renters. The way
52:06
52 minutes, 6 seconds
we use CPUs — we live in a world counted by seconds. The way we rent CPUs in the cloud,
52:15
52 minutes, 15 seconds
the more CPU cores you have, the more you can rent. The economics and the use case of the
52:24
52 minutes, 24 seconds
old CPU are fundamentally different from agents. Agents are impatient. They don’t live in a world
52:35
52 minutes, 35 seconds
that is counted in seconds. They live in a world counted in nanoseconds. When an agent uses a tool,
52:42
52 minutes, 42 seconds
it wants the response time to be as fast as possible. When it accesses a database, it has
52:47
52 minutes, 47 seconds
to come back as soon as possible. Every moment that the agent is waiting keeps it from going
52:55
52 minutes, 55 seconds
to the next step, the next step, the next step. It is vital that we make the CPUs as low-latency
53:04
53 minutes, 4 seconds
as possible, as interactive as possible. So we created Vera CPU for the age of AI.
53:12
53 minutes, 12 seconds
Now inside our system, it’s used in three different ways. The first way, of course,
53:17
53 minutes, 17 seconds
is within Vera Rubin. For thinking — and inside the Vera Rubin rack, there are already two
53:26
53 minutes, 26 seconds
CPUs. As you know, we are building and selling millions of Vera Rubins. We have sold millions
53:36
53 minutes, 36 seconds
of Grace Blackwells. And NVIDIA already is one of the largest CPU makers in the world.
53:42
53 minutes, 42 seconds
In the Vera Rubin rack there are two CPUs. One for orchestrating and managing the GPUs,
53:51
53 minutes, 51 seconds
managing the KV cache, dealing with all of the software that runs in the rack. We also have
53:59
53 minutes, 59 seconds
the Grace BlueField that is used for security and isolation. The Vera compute is used for
54:08
54 minutes, 8 seconds
the harness, the orchestration of the AI models, tool use, accessing the database. And the data
54:16
54 minutes, 16 seconds
servers are right here — Vera BlueField — the fastest storage servers the world has ever made.
54:26
54 minutes, 26 seconds
And the reason why this is so vital is because agents are accessing memory so incredibly fast.
54:35
54 minutes, 35 seconds
These systems — the storage server and the CPUs — are in the critical path of the most
54:44
54 minutes, 44 seconds
expensive part of the data center. This is the most expensive for a good reason. The
54:51
54 minutes, 51 seconds
economics of the AI factory are tokens. And the tokens are created here. And so, of course, you
55:01
55 minutes, 1 second
want to manufacture and generate as many tokens as possible. This is where you put all of your
55:06
55 minutes, 6 seconds
economics, and this has to not be in the way. And so Vera CPU has great requirements on the
55:15
55 minutes, 15 seconds
CPU architecture, which is the reason why we built a brand new architecture from the
55:20
55 minutes, 20 seconds
ground up. A CPU the world has never seen before. We call it Vera. This is CPU for
55:29
55 minutes, 29 seconds
agents. All the CPUs of the past, we built for humans. This CPU is built for agents.
55:37
55 minutes, 37 seconds
There are four things to keep in mind — the four takeaways. The first takeaway
55:43
55 minutes, 43 seconds
is that the instructions per clock of Vera has to be incredibly good,
55:49
55 minutes, 49 seconds
because we need the latency to be short. We need the processing time — single-threaded performance,
55:56
55 minutes, 56 seconds
not throughput — single-threaded performance has to be world class. Absolutely the best. Which is
56:03
56 minutes, 3 seconds
the reason why the IPC, the instructions per clock of Vera, is so high. It’s the
56:09
56 minutes, 9 seconds
highest in the world. 10 instructions fetched, decoded, and executed per clock. Number one.
56:17
56 minutes, 17 seconds
Number two, the bandwidth necessary to move data in and out of the CPU has to
56:24
56 minutes, 24 seconds
be utterly world class — bandwidth per core. The third is just bandwidth, period. We’re
56:34
56 minutes, 34 seconds
moving data — and remember, I said earlier, agentic systems are fundamentally disaggregated
56:41
56 minutes, 41 seconds
and distributed. Disaggregated and distributed. When computing is disaggregated and distributed,
56:49
56 minutes, 49 seconds
networking becomes the problem. Therefore, we have to move the data around as fast as
56:55
56 minutes, 55 seconds
possible — between the CPU cores, between the CPU and the storage, the CPU and the GPU. The
57:02
57 minutes, 2 seconds
bandwidth around the system, and inside the CPU core, has to be utterly world class.
57:09
57 minutes, 9 seconds
This is the first CPU that’s been built in a long time that is literally at radical limits,
57:15
57 minutes, 15 seconds
with a fabric that connects all of the CPU cores at the speed of light — 3.6 terabytes
57:22
57 minutes, 22 seconds
per second. No trip boundary crossings, because all of the CPU cores are talking to each other
57:34
57 minutes, 34 seconds
with extremely high bandwidth. They’re not rented core by core. They’re all working together. The
57:42
57 minutes, 42 seconds
cross-sectional bandwidth of Vera is off the charts. It’s the first CPU to use PCIe Gen 6.
57:49
57 minutes, 49 seconds
It is also the first to use LPDDR5 with 1.2 terabytes per second — two to three times the
57:59
57 minutes, 59 seconds
bandwidth of the highest-performance CPUs in the market. Three times the bandwidth on the inside.
58:06
58 minutes, 6 seconds
The bandwidth per core, and the bandwidth overall, is world class. Now, remember — the
58:16
58 minutes, 16 seconds
number of CPUs is going to be quite high. And the reason for that is very simple. We created
58:24
58 minutes, 24 seconds
CPUs for humans in the past, and humans — there are only a billion of us. There will be billions
58:33
58 minutes, 33 seconds
of agents. And these agents are going to be using the CPUs with very little patience, because the
58:41
58 minutes, 41 seconds
cost of the GPU that they sit next to is too high, too valuable, too precious. Therefore,
58:50
58 minutes, 50 seconds
these CPUs are going to be both performant, but they also have to be extremely energy-efficient,
58:59
58 minutes, 59 seconds
so that we can pack as much CPU as we can into the factory without taking away power from the token
59:07
59 minutes, 7 seconds
generation, which we know is how we make money. These four properties — instructions per clock,
59:13
59 minutes, 13 seconds
or single-threaded performance; bandwidth per core; the total bandwidth around the chip and
59:21
59 minutes, 21 seconds
inside the chip; and energy efficiency — define Vera. It is absolutely world class. When you
59:28
59 minutes, 28 seconds
compare it to the highest-performance x86, it is just off the charts. When you compare it in real,
59:35
59 minutes, 35 seconds
single-threaded performance — real performance — it’s off the charts. It is incredible to be
59:44
59 minutes, 44 seconds
able to deliver a 5% improvement on CPUs. It is incredible to be able to deliver 10%. But this
59:51
59 minutes, 51 seconds
kind of performance speedup is just unheard of. This is NVIDIA Vera. What do you think?
1:00:05
1 hour, 5 seconds
Let’s take a look. ________________ [VIDEO NARRATION] Agentic AI changes the role
1:00:10
1 hour, 10 seconds
of the CPU. The CPU is now the conductor, and the GPU is the orchestra. Traditional CPUs were
1:00:17
1 hour, 17 seconds
built for a different era — maximizing cores per socket, slicing them up, virtualizing, renting by
1:00:24
1 hour, 24 seconds
the hour. In the age of agents, the CPU is now a bottleneck to GPU utilization, directly affecting
1:00:32
1 hour, 32 seconds
token throughput, latency, and user experience. NVIDIA Vera is the CPU built for the agentic
1:00:39
1 hour, 39 seconds
loop — combining NVIDIA’s custom data center CPU core with a scalable coherency fabric,
1:00:45
1 hour, 45 seconds
for the right balance of performance, cores, and bandwidth to maximize AI factory health. At the heart of Vera is the NVIDIA Olympus core, built for modern data
1:00:55
1 hour, 55 seconds
center workloads — branch-heavy Python runtimes, tool calls, and sandbox code execution. Each core
1:01:03
1 hour, 1 minute, 3 seconds
is tuned for throughput. A neural branch predictor evaluating two taken branches per cycle. A 10-wide
1:01:10
1 hour, 1 minute, 10 seconds
decode engine brings in more work each cycle. A large out-of-order engine keeps instructions
1:01:15
1 hour, 1 minute, 15 seconds
moving. Advanced prefetchers with a novel graph engine anticipating the next data fetch.
1:01:21
1 hour, 1 minute, 21 seconds
But fast cores only matter when data arrives correctly and on time. Vera is
1:01:27
1 hour, 1 minute, 27 seconds
the first CPU to use LPDDR5X memory while correcting multiple errors simultaneously
1:01:34
1 hour, 1 minute, 34 seconds
without compromising bandwidth. Vera achieves 40% lower peak memory latency
1:01:39
1 hour, 1 minute, 39 seconds
versus x86, keeping cores fed on time through retrieval, analytics, and sandbox execution.
1:01:47
1 hour, 1 minute, 47 seconds
NVIDIA’s second-generation scalable coherence fabric unifies all 88 Olympus cores on a
1:01:53
1 hour, 1 minute, 53 seconds
monolithic mesh. Separate dies for memory and cores are not split across chiplets, enabling 50%
1:02:02
1 hour, 2 minutes, 2 seconds
faster core-to-core communication than traditional CPUs. And memory-coherent NVLink chip-to-chip
1:02:09
1 hour, 2 minutes, 9 seconds
connects GPUs directly to the CPU. Beyond GPUs, NVLink chip-to-chip can scale Vera up to multiple
1:02:17
1 hour, 2 minutes, 17 seconds
sockets, enabling massive bandwidth between CPUs. Vera delivers 1.8 times the agentic sandbox
1:02:25
1 hour, 2 minutes, 25 seconds
performance of x86 CPUs. Standalone, Vera racks run agent sandboxes, tools, code,
1:02:32
1 hour, 2 minutes, 32 seconds
and data pipelines. Tightly coupled to Rubin GPUs, Vera keeps accelerated workloads moving.
1:02:39
1 hour, 2 minutes, 39 seconds
NVIDIA Vera BlueField-4 STX — powers context, memory, and AI storage. Compute, network, storage.
1:02:49
1 hour, 2 minutes, 49 seconds
Vera is the CPU for the age of agents. ________________
1:03:01
1 hour, 3 minutes, 1 second
[JENSEN] This is going to be our new major
1:03:03
1 hour, 3 minutes, 3 seconds
growth driver. The reviews are already coming out, and it’s pretty good. Pretty good stuff.
1:03:18
1 hour, 3 minutes, 18 seconds
Now, remember — Grace and Vera are also the most highly qualified CPUs in the world of AI,
1:03:30
1 hour, 3 minutes, 30 seconds
because every single data center, every single cloud, every single enterprise,
1:03:35
1 hour, 3 minutes, 35 seconds
every company that works with NVIDIA on AI has already qualified Grace. The entire software
1:03:43
1 hour, 3 minutes, 43 seconds
stack has already been optimized for Grace. Every company will be qualifying Vera. Vera will be the
1:03:50
1 hour, 3 minutes, 50 seconds
most optimized agentic CPU in the world, simply because it’s going to go with Vera Rubin, simply
1:03:57
1 hour, 3 minutes, 57 seconds
because we made the big architecture switch. In fact, during the Grace Blackwell transition,
1:04:02
1 hour, 4 minutes, 2 seconds
the biggest risk was going from external CPU x86 into Grace Blackwell. That transition
1:04:11
1 hour, 4 minutes, 11 seconds
was extremely challenging. But we did it with incredible execution. Now Grace is literally
1:04:19
1 hour, 4 minutes, 19 seconds
synonymous with Grace Blackwell. When people say Blackwell, they say Grace Blackwell, because it is
1:04:25
1 hour, 4 minutes, 25 seconds
now utterly everywhere. Every company’s software stack has been optimized for it. Everybody’s security stack has been optimized for it. And now here comes Vera. I’m super excited about that.
1:04:36
1 hour, 4 minutes, 36 seconds
Now, look at some of the performance numbers. Speedups are one thing. It is extremely hard
1:04:42
1 hour, 4 minutes, 42 seconds
to speed up SQL. SQL — the most famous domain-specific language that has ever
1:04:54
1 hour, 4 minutes, 54 seconds
been created. Before CUDA, it was SQL. Before OpenGL, there was SQL. Invented by IBM. Today,
1:05:05
1 hour, 5 minutes, 5 seconds
it is the structured database language of the planet. Everybody uses SQL. This is
1:05:11
1 hour, 5 minutes, 11 seconds
SQL running three times faster — not 10% faster, not 25% faster. Three times faster. Incredible.
1:05:25
1 hour, 5 minutes, 25 seconds
The next one is real-time stream processing. Remember, your AI is going to be not just reading
1:05:32
1 hour, 5 minutes, 32 seconds
documents. Your AI is going to be watching for telemetry — especially inside a factory,
1:05:38
1 hour, 5 minutes, 38 seconds
inside a stock exchange. You’re going to be looking for telemetry continuously.
1:05:44
1 hour, 5 minutes, 44 seconds
The burst of data that’s coming in goes into a CPU. This is Vera CPU
1:05:50
1 hour, 5 minutes, 50 seconds
running real-time stream processing for the New York Stock Exchange. Lynn Martin, the president of the New York Stock Exchange, has been so gracious to partner with us.
1:06:00
1 hour, 6 minutes
This system runs all over the world, in real time — real-time stream processing. Vera CPU:
1:06:06
1 hour, 6 minutes, 6 seconds
six times faster. All because of the bandwidth, the single-threaded instruction execution, the
1:06:14
1 hour, 6 minutes, 14 seconds
bandwidth inside between the cores, the bandwidth outside. Vera is completely revolutionary.
1:06:28
1 hour, 6 minutes, 28 seconds
You know, talking about x-factors is something you do when you’re talking about GPUs. It is
1:06:33
1 hour, 6 minutes, 33 seconds
quite rare that somebody talks about x-factors on real workloads associated with a CPU. So I’m
1:06:42
1 hour, 6 minutes, 42 seconds
so proud of the team. You guys did such a great job. We have an extraordinary roadmap coming.
1:06:48
1 hour, 6 minutes, 48 seconds
And what’s really exciting is almost everybody is supporting Vera. They’re as excited as we
1:06:55
1 hour, 6 minutes, 55 seconds
are. This is Vera opening up a brand new market. Agents is a new workload. We built
1:07:06
1 hour, 7 minutes, 6 seconds
CPUs for humans in the past. We need CPUs for agents, for agentic systems. The properties are
1:07:14
1 hour, 7 minutes, 14 seconds
different. Why would the CPU be the same? We are building millions and millions of Veras.
1:07:24
1 hour, 7 minutes, 24 seconds
And to go to market with us — Taiwan’s ODMs and computer makers, all the OEMs — you can see the
1:07:33
1 hour, 7 minutes, 33 seconds
early adopters. The early adopters are the agentic companies. This is the beginning of a new market,
1:07:41
1 hour, 7 minutes, 41 seconds
a market that never existed before. It’s not going to take away from the old markets,
1:07:46
1 hour, 7 minutes, 46 seconds
but this is a new market: CPU for agents. And this market will surely be larger than the last,
1:07:57
1 hour, 7 minutes, 57 seconds
because there will be a lot more agents than there are people, and the agents are very impatient.
1:08:03
1 hour, 8 minutes, 3 seconds
So, NVIDIA Vera CPU. Thank you. This is the most important slide, really. This
1:08:16
1 hour, 8 minutes, 16 seconds
is the takeaway. The takeaway here is that this is the application pattern — the computing pattern of
1:08:23
1 hour, 8 minutes, 23 seconds
the next decade. Agents, harnesses, orchestrating large language models. Every company will run
1:08:34
1 hour, 8 minutes, 34 seconds
1:08:43
1 hour, 8 minutes, 43 seconds
Every company will see that agents need their own operating system. Every company is asking:
1:08:51
1 hour, 8 minutes, 51 seconds
how do we run agents safely? How do we build agents for our own workloads?
1:08:58
1 hour, 8 minutes, 58 seconds
And so, we have the NVIDIA Agent Toolkit for Enterprise AI. You’ve seen me build
1:09:06
1 hour, 9 minutes, 6 seconds
this in plain sight. Almost everything that NVIDIA does — at every GTC — if you go back
1:09:11
1 hour, 9 minutes, 11 seconds
and look at my GTC from five years ago or 10 years ago, you will see today. This,
1:09:17
1 hour, 9 minutes, 17 seconds
you’ve seen me talking about for several years, because we’ve been building for this moment.
1:09:23
1 hour, 9 minutes, 23 seconds
There are four things that companies need in order to build agents as a service,
1:09:29
1 hour, 9 minutes, 29 seconds
or to build agents to operate. The first thing you need is models — of course, large language models,
1:09:38
1 hour, 9 minutes, 38 seconds
the smarter the better, the cheaper the better, the faster the better. The second is you need
1:09:44
1 hour, 9 minutes, 44 seconds
a harness to orchestrate the whole thing. The third — these models want to use tools,
1:09:52
1 hour, 9 minutes, 52 seconds
and these tools come with skills. And I showed you CUDA-X libraries — those are
1:09:57
1 hour, 9 minutes, 57 seconds
going to be amazing tools for the agents in the future. And then lastly, you need a runtime. You
1:10:05
1 hour, 10 minutes, 5 seconds
need the operating system that holds it all together. This is the NVIDIA Agent Toolkit.
1:10:14
1 hour, 10 minutes, 14 seconds
It includes models that you can modify — various world-class open models, and I’ll
1:10:21
1 hour, 10 minutes, 21 seconds
show you more. You can run agents from anybody. You can run Claude Code — an incredible agent.
1:10:29
1 hour, 10 minutes, 29 seconds
Codex — incredible agent. You could run it inside this harness called NVIDIA OpenShell, which is
1:10:35
1 hour, 10 minutes, 35 seconds
highly secure for inside the enterprise. The shell protects the agent, keeps it grounded in security
1:10:44
1 hour, 10 minutes, 44 seconds
policies. Privacy is protected, its rights and privileges are given, its identity is protected.
1:10:53
1 hour, 10 minutes, 53 seconds
And so this OpenShell is being adopted all over the world, and it is open source. You can see so
1:11:00
1 hour, 11 minutes
many companies adopting it — Red Hat, Canonical, Microsoft — it’s going to be adopted everywhere.
1:11:07
1 hour, 11 minutes, 7 seconds
This is the runtime. And this runtime is fully optimized for the NVIDIA AI platform, which is
1:11:16
1 hour, 11 minutes, 16 seconds
everywhere. So you can run OpenShell in any cloud, on-prem, and even on device. You have tools and
1:11:26
1 hour, 11 minutes, 26 seconds
libraries that agents can use. You have models that you can modify or use as-is. Or you have
1:11:34
1 hour, 11 minutes, 34 seconds
agents — OpenClaw, Hermes — another incredible agentic harness. These agentic harnesses can
1:11:46
1 hour, 11 minutes, 46 seconds
now run on-prem or for you anywhere.
So, four things, and this represents the operating system of the modern enterprise. Now, how do we use this? One of my favorite use cases of agents is chip designers. It is the single most important thing that NVIDIA does. And so, of course, we have to partner with Cadence to build a super agent — a chip design super agent.
1:12:16
1 hour, 12 minutes, 16 seconds
It is orchestrated by Codex or Claude Code. It has RTL and architecture diagrams, or schematics,
1:12:25
1 hour, 12 minutes, 25 seconds
or specifications as input, and whatever you need to fix. And together, we created some
1:12:33
1 hour, 12 minutes, 33 seconds
super agents that are optimized for the NVIDIA platform, with Nemotron. Let’s take a look.
1:12:42
1 hour, 12 minutes, 42 seconds
Chartering 5 Layer AI Agency - integrating exponential intergenerational multipliers of trusted human relationship systems through community scaling apps
AsiaAI.docx where & how 2/3 human brains are celebrating AI livelihoods
====
lelated US AI reports:
chris.macrae@yahoo.co.uk :help celebrate library of INTELLIGENCE multipliers: -system map
views on whether AGI exists
- how close are google aws or huawei to nvidia
2025REPORT-ER: Entrepreneurial Revolution est 1976; Neumann Intelligence Unit at The Economist since 1951. Norman Macrae's & friends 75 year mediation of engineers of computing & autonomous machines has reached overtime: Big Brother vs Little Sister !?
Overtime help ed weekly quizzes on Gemini of Musk & Top 10 AI brains until us election nov 2028
unaiwho.docx version 6/6/22 hunt for 100 helping guterres most with UN2.0
RSVP chris.macrae@yahoo.co.uk
Prep for UNSUMMITFUTURE.com
JOIN SEARCH FOR UNDER 30s MOST MASSIVE COLLABS FOR HUMAN SUSTAINABILITY
1 Jensen Huang 2 Demis Hassabis 3 Dei-Fei Li 4 King Charles
5 Bezos Earth (10 bn) 6 Bloomberg JohnsHopkins cbestAI.docx 7 Banga
8 Maurice Chang 9 Mr & Mrs Jerry Yang 10 Mr & Mrs Joseph Tsai 11 Musk
12 Fazle Abed 13 Ms & Mr Steve Jobs 14 Melinda Gates 15 BJ King 16 Benioff
17 Naomi Osaka 18 Jap Emperor Family 19 Akio Morita 20 Mayor Koike
The Economist 1982 why not Silicon AI Valley Everywhere 21 Founder Sequoia 22 Mr/Mrs Anne Doerr 23 Condi Rice
23 MS & Mr Filo 24 Horvitz 25 Michael Littman NSF 26 Romano Prodi 27 Andrew Ng 29 Lila Ibrahim 28 Daphne Koller
30 Mayo Son 31 Li Ka Shing 32 Lee Kuan Yew 33 Lisa Su 34 ARM 36 Priscilla Chan
38 Agnelli Family 35 Ms Tan & Mr Joe White
37 Yann Lecun 39 Dutch Royal family 40 Romano Prodi
41 Kramer 42 Tirole 43 Rachel Glennerster 44 Tata 45 Manmohan Singh 46 Nilekani 47 James Grant 48 JimKim, 49 Guterres
50 attenborough 51 Gandhi 52 Freud 53 St Theresa 54 Montessori 55 Sunita Gandhu,56 paulo freire 57 Marshall Mcluhan58 Andrew Sreer 59 Lauren Sanchez, 60 David Zapolski
61 Harris 62 Chips Act Raimundo 63 oiv Newsom. 64 Arati Prab hakarm,65 Jennifer Doudna CrispR, 66 Oren Etsioni,67 Robert Reisch,68 Jim Srreyer 69 Sheika Moza
- 3/21/22 HAPPY 50th Birthday TO WORLD'S MOST SUSTAINABLE ECONOMY- ASIAN WOMEN SUPERVILLAGE
Since gaining my MA statistics Cambridge DAMTP 1973 (Corpus Christi College) my special sibject has been community building networks- these are the 6 most exciting collaboration opportunities my life has been privileged to map - the first two evolved as grassroots person to person networks before 1996 in tropical Asian places where village women had no access to electricity grids nor phones- then came mobile and solar entrepreneurial revolutions!!
COLLAB platforms of livesmatter communities to mediate public and private -poorest village mothers empowering end of poverty 5.1 5.2 5.3 5.4 5.5 5.6
4 livelihood edu for all
3 last mile health services 3.1 3,2 3.3 3.4 3.5 3.6
last mile nutrition 2.1 2.2 2.3 2.4 2.5 2,6
banking for all workers 1.1 1.2 1.3 1.4 1.5 1.6
NEWS FROM LIBRARY NORMAN MACRAE -latest publication 2021 translation into japanese biography of von neumann:
Below: neat German catalogue (about half of dad's signed works) but expensive -interesting to see how Germans selected the parts they like over time: eg omitted 1962 Consider Japan The Economist
feel free to ask if free versions are available
0 The coming entrepreneurial revolution : a survey - In: The economist 261 (1976), pp. 41-65 cited 105
. we scots are less than 4/1000 of the worlds and 3/4 are Diaspora - immigrants in others countries. Since 2008 I have been celebrating Bangladesh Women Empowerment solutions wth NY graduates. Now I want to host love each others events in new york starting this week with hong kong-contact me if we can celebrate anoither countries winm-wins with new yorkers
mapping OTHER ECONOMIES:
50 SMALLEST ISLAND NATIONS
TWO Macroeconomies FROM SIXTH OF PEOPLE WHO ARE WHITE & war-prone
ADemocratic
Russian
=============
From 60%+ people =Asian Supercity (60TH YEAR OF ECONOMIST REPORTING - SEE CONSIDER JAPAN1962)
Far South - eg African, Latin Am, Australasia
Earth's other economies : Arctic, Antarctic, Dessert, Rainforest
===========
In addition to how the 5 primary sdgs1-5 are gravitated we see 6 transformation factors as most critical to sustainability of 2020-2025-2030
Xfactors to 2030 Xclimate XAI Xinfra Xyouth Wwomen Xpoor chris.macrae@yahoo.co.uk (scot currently in washington DC)- in 1984 i co-authored 2025 report with dad norman.
|
Asia Rising Surveys
Entrepreneurial Revolution -would endgame of one 40-year generations of applying Industrial Revolution 3,4 lead to sustainability of extinction 1972's Next 40 Years ;1976's Coming Entrepreneurial Revolution; 12 week leaders debate 1982's We're All Intrapreneurial Now
|
The Economist had been founded in 1843" marking one of 6 exponential timeframes "Future Histores"
IN ASSOCIATION WITH ADAMSMITH.app :
we offer worldwide mapping view points from
and these viewpoints:
40 years ago -early 1980s when we first framed 2025 report;
from 1960s when 100 times more tech per decade was due to compound industrial revolutions 3,4
1945 birth of UN
1843 when the economist was founded
1760s - adam smithian 2 views : last of pre-engineering era; first 16 years of engineering ra including america's declaration of independence- in essence this meant that to 1914 continental scaling of engineeriing would be separate new world <.old world
IF we 8 billion earthlings of the 2020s are to celebrate collaboration escapes from extinction, the knowhow of the billion asian poorest women networks will be invaluable -
in mathematically connected ways so will the stories of diaspora scots and the greatest mathematicians ever home schooled -central european jewish teens who emigrated eg Neumann , Einstein ... to USA 2nd quarter of the 20th century; it is on such diversity that entrepreneurial revolution diaries have been shaped
EconomistPOOR.com : Dad was born in the USSR in 1923 - his dad served in British Embassies. Dad's curiosity enjoyed the opposite of a standard examined education. From 11+ Norman observed results of domination of humans by mad white men - Stalin from being in British Embassy in Moscow to 1936; Hitler in Embassy of last Adriatic port used by Jews to escape Hitler. Then dad spent his last days as a teen in allied bomber command navigating airplanes stationed at modernday Myanmar. Surviving thanks to the Americas dad was in Keynes last class where he was taught that only a handful of system designers control what futures are possible. EconomistScotland.com AbedMooc.com
To help mediate such, question every world eventwith optimistic rationalism, my father's 2000 articles at The Economist interpret all sorts of future spins. After his 15th year he was permitted one signed survey a year. In the mid 1950s he had met John Von Neumann whom he become biographer to , and was the only journalist at Messina's's birth of EU. == If you only have time for one download this one page tour of COLLABorations composed by Fazle Abed and networked by billion poorest village women offers clues to sustainability from the ground up like no white ruler has ever felt or morally audited. by London Scot James Wilson. Could Queen Victoria change empire fro slavemaking to commonwealth? Some say Victoria liked the challenge James set her, others that she gave him a poison pill assignment. Thus James arrived in Calcutta 1860 with the Queens permission to charter a bank by and for Indian people. Within 9 months he died of diarrhea. 75 years later Calcutta was where the Young Fazle Abed grew up - his family accounted for some of the biggest traders. Only to be partitioned back at age 11 to his family's home region in the far north east of what had been British Raj India but was now to be ruled by Pakistan for 25 years. Age 18 Abed made the trek to Glasgow University to study naval engineering.
1943 marked centenary autobio of The Economist and my teenage dad Norman prepping to be navigator allied bomber command Burma Campaign -thanks to US dad survived, finished in last class of Keynes. before starting 5 decades at The Economist; after 15 years he was allowed to sign one survey a year starting in 1962 with the scoop that Japan (Korea S, Taiwan soon hk singapore) had found development mp0de;s for all Asian to rise. Rural Keynes could end village poverty & starvation; supercity win-win trades could celebrate Neumanns gift of 100 times more tech per decade (see macrae bio of von neumann)
Since 1960 the legacy of von neumann means ever decade multiplies 100 times more micro-technology- an unprecedented time for better or worse of all earthdwellers; 2025 timelined and mapped innovation exponentials - education, health, go green etc - (opportunities threats) to celebrating sustainability generation by 2025; dad parted from earth 2010; since then 2 journals by adam smith scholars out of Glasgow where engines began in 1760- Social Business; New Economics have invited academic worlds and young graduates to question where the human race is going - after 30 business trips to wealthier parts of Asia, through 2010s I have mainly sherpa's young journalist to Bangladesh - we are filing 50 years of cases on women empowerment at these web sites AbedMOOC.com FazleAbed.com EconomistPoor.com EconomistUN.com WorldRecordjobs.com Economistwomen.com Economistyouth.com EconomistDiary.com UNsummitfuture.com - in my view how a billion asian women linked together to end extreme poverty across continental asia is the greatest and happiest miracle anyone can take notes on - please note the rest of this column does not reflect my current maps of how or where the younger half of the world need to linkin to be the first sdg generation......its more like an old scrap book
how do humans design futures?-in the 2020s decade of the sdgs – this question has never had more urgency. to be or not to be/ – ref to lessons of deming or keynes, or glasgow university alumni smith and 200 years of hi-trust economics mapmaking later fazle abed - we now know how-a man made system is defined by one goal uniting generations- a system multiplies connected peoples work and demands either accelerating progress to its goal or collapsing - sir fazle abed died dec 2020 - so who are his most active scholars climate adaptability where cop26 november will be a great chance to renuite with 260 years of adam smith and james watts purposes t end poverty-specifically we interpret sdg 1 as meaning next girl or boy born has fair chance at free happy an productive life as we seek to make any community a child is born into a thriving space to grow up between discover of new worlds in 1500 and 1945 systems got worse and worse on the goal eg processes like slavery emerged- and ultimately the world was designed around a handful of big empires and often only the most powerful men in those empires. 4 amazing human-tech systems were invented to start massive use by 1960 borlaug agriculture and related solutions every poorest village (2/3people still had no access to electricity) could action learn person to person- deming engineering whose goal was zero defects by helping workers humanize machines- this could even allowed thousands of small suppliers to be best at one part in machines assembled from all those parts) – although americans invented these solution asia most needed them and joyfully became world class at them- up to 2 billion people were helped to end poverty through sharing this knowhow- unlike consuming up things actionable knowhow multiplies value in use when it links through every community that needs it the other two technologies space and media and satellite telecoms, and digital analytic power looked promising- by 1965 alumni of moore promised to multiply 100 fold efficiency of these core tech each decade to 2030- that would be a trillion tmes moore than was needed to land on the moon in 1960s. you might think this tech could improve race to end poverty- and initially it did but by 1990 it was designed around the long term goal of making 10 men richer than 40% poorest- these men also got involved in complex vested interests so that the vast majority of politicians in brussels and dc backed the big get bigger - often they used fake media to hide what they were doing to climate and other stuff that a world trebling in population size d\ - we the 3 generations children parents grandparents have until 2030 to design new system orbits gravitated around goal 1 and navigating the un's other 17 goals do you want to help/ 8 cities we spend most time helping students exchange sustainability solutions 2018-2019 BR0 Beijing Hangzhou:
Girls world maps begin at B01 good news reporting with fazleabed.com valuetrue.com and womenuni.com
.==========
online library of norman macrae--
==========
Ma 2 Ali Financial
health catalogue; energy catalogue
Keynes: 2025now - jobs Creating Gen
.
how poorest women in world build
A01 BRAC health system,
A02 BRAC education system,
A03 BRAC banking system
K01 Twin Health System - Haiti& Boston
Past events EconomistDiary.com
include 15th annual spring collaboration cafe new york - 2022 was withsister city hong kong designers of metaverse for beeings.app
© 2026 Created by chris macrae.
Powered by
Chris AI Macrae MA DAMTP CantabAuthor
Prediction AI must urgently change education - how teachers and students spend their time- but this doesn't have to mean less teachers
You just give every one of those teachers a personalized curriculum for every student in the room. It makes them all bionic and leads to a lot more. Every single student will be assisted by AI,
but every single student will need great teachers. Yeah. Yeah. Amazing.
Chris AI Macrae MA DAMTP CantabAuthor
where is AI have a true impact on the health care system?
Jensen : There's several several areas that Nvidia is involved in in um in healthcare. 1) AI physics and that's AI biology using AI to understand represent behavior biologically
that's very important in drug discovery. There's second which is AI agents and that's where the assistance and helping diagnosis and things like that. Two native ai companies : Open evidence and Hippocratic are really good examples.
. A third part that we're in involved in is physical AI.to understand the properties of the laws of physics and that's used for robotic surgery Every single instrument whether it's ultrasound or CT or whatever we interact with in a hospital in the future will be agentic.
IE open claw in a safe version will be inside every single instrument. Footnote Jensen listed 100 native ai companies at GTC 2026 https://www.economistdiary.com/
Jensen Huang: Nvidia's Future, Physical AI, Rise of the Agent, Inference Explosion, AI PR Crisis
(0:00) Jensen Huang joins the show! (0:26) Acquiring Groq and the inference explosion (8:53) Decision making at the world's most valuable company (10:47) Physical AI's $50T market, OpenClaw's future, the new operating system for modern AI computing...
Chris AI Macrae MA DAMTP CantabAuthor
when are we to see humanoid robots in the home, and is China ahead on thiis?
Jensen, actually the US invented the industry largely.
you could argue we got into it too soon and We got tired um about five years before the enabling technology appeared.. But it's here now. From the point of high functioning existence proof, to reasonable products technology never takes more than a couple two three cycles. And so a couple
that will likely be three years to 5 years away before we have robots all over the place..
Yes I think China is formidable in humanoids and the reason for that is because their micro electronics,
motors, their rare earth, their magnets, which are foundational to robotics, they are the world's best. And so in a lot of ways, our robotics industry relies deeply on their ecosystem and their supply chain. But this is obviously moving very quickly. . (interviewer some like Elon seems to think we're going to have one robot for every human. 7 billion for 7 billion, 8 billion for 8 billion.)
Jensen Well, I'm hoping more
Jensen Huang: Nvidia's Future, Physical AI, Rise of the Agent, Inference Explosion, AI PR Crisis
(0:00) Jensen Huang joins the show! (0:26) Acquiring Groq and the inference explosion (8:53) Decision making at the world's most valuable company (10:47) Physical AI's $50T market, OpenClaw's future, the new operating system for modern AI computing...
Chris AI Macrae MA DAMTP CantabAuthor
to jensen - you've made massive progress in self-driving cars and hundreds of partners - how do you navigate that & your big new announcement including BYD and ever deeper partnerships as different as mercedes, and uber and 11+ years platform with Tesla?
Jensen Firstly we are the only full stack AI company serving platforms needed for self driving cars- and We believe that everything that moves will be autonomous completely or partly
someday. Moreover, what we don't do is build self-driving cars,
but we want to enable every car company in the world to build self-driving cars. And so, we built all three computers, :the training computer, the simulation computer, the evaluation computer,
as well as the car computer. We develop the world's safest driving operating system. We also created the world's first reasoning autonomous vehicle so that it could decompose complicated scenarios into simpler scenarios that it knows how to navigate through just like all deep reasoning systems. Alpommyo Reasoning system is achieving incredible results. We open this but offer vertical optimization. We horizontally innovate and let everybody decide. what Do you want
Jensen Huang: Nvidia's Future, Physical AI, Rise of the Agent, Inference Explosion, AI PR Crisis
(0:00) Jensen Huang joins the show! (0:26) Acquiring Groq and the inference explosion (8:53) Decision making at the world's most valuable company (10:47) Physical AI's $50T market, OpenClaw's future, the new operating system for modern AI computing...
Chris AI Macrae MA DAMTP CantabAuthor
Jensen-Nobody wants the US to win global AI race more than you,? But a year ago, Biden era was anti- American diffusion of AI around world. So into the new administration. - where are we in terms of global diffusion of US AI technology?
Jensen Well, first, President Trump wants American industry to lead... to spread American technology around the world. ..United States to be the wealthiest country in the world. He wants all of that. Biden admin: Nvidia gave up a 95% market share in the second largest market..
President Trump wants us to get back. And first, to get
license licensed for the companies that we're going to be able to sell to. We've got many companies who have requested for licenses. We've applied for licenses for them and we've got approved licenses from sec secretary lutnik. we've informed the Chinese companies so we're in the process of cranking up