Nvidia has shifted its focus from designing individual chips to building entire rack-scale systems. This shift is driven by the fact that modern AI problems are too large for a single computer. When thousands of computers work together, every component becomes a potential bottleneck. If computation speeds up but networking remains slow, the overall system gains are limited. This is the core challenge of Amdahl's Law in distributed computing.

The reason why extreme co-design is necessary is because the problem no longer fits inside one computer to be accelerated by one GPU. You would like to go faster than the number of computers that you add. Then all of a sudden you have to take the algorithm, you have to break it up, you have to shard the pipeline, you have to shard the data, you have to shard the model.

To manage this complexity, Jensen structured his organization to mirror the product. He has 60 direct reports and avoids one-on-one meetings. Instead, problems are discussed in large groups where experts in memory, optics, and cooling can all hear each other. This transparency prevents a design choice in one area from negatively impacting another. It ensures the entire stack is optimized in unison.

When you're designing a company, you should first think about what is it that you want the company to produce. The architecture of the company should reflect the environment by which it exists. It almost directly says what you should do with the organization.

This organizational approach allows for constant cross-discipline feedback. If a specialist proposes a cooling solution that does not work for power distribution, other experts can immediately flag the issue. This creates a collaborative environment where everyone is expected to contribute to the high-level system design.

Nvidia began as a company focused on specialized accelerators. While specialization allows for extreme optimization, it also limits market reach and the ability to fund research and development. Jensen explains that there is a natural tension between being a great specialist and being a general computing company. The more a company specializes, the less capacity it has for overall computing. Nvidia had to find a narrow path to expand its capabilities without losing its core strengths.

The journey toward general computing happened in small steps. First, the company invented the programmable pixel shader. Later, they added IEEE compatible FP32 support. This was a massive shift because it allowed researchers to use GPUs for tasks previously reserved for CPUs. These steps eventually led to the creation of CUDA, a parallel computing platform and programming model.

To make CUDA successful, Nvidia had to solve the problem of adoption. Developers only join a platform if it has a large install base. Jensen points to the X86 architecture as an example. Even if an architecture is not the most elegant, it can dominate if it is everywhere. Many beautiful architectures failed because they lacked a large enough user base.

The install base is in fact the single most important part of an architecture. We ought to put CUDA on GeForce and put it into every single PC, whether customers use it or not, and use it as a starting point of cultivating our installed base.

The most difficult strategic decision was putting CUDA on every GeForce GPU. At the time, GeForce was a consumer product for gamers who did not necessarily need CUDA. This move increased the cost of each chip by 50 percent. Since Nvidia was only a 35 percent gross margin company, this decision nearly destroyed its profits. The company's market value dropped from around 8 billion dollars to 1.5 billion dollars. Jensen describes this as an existential threat. However, the move allowed researchers and students to discover CUDA on the gaming PCs they already owned. This created the foundation for the deep learning revolution. GeForce provided the reach that CUDA needed to become a standard.

Jensen approaches leadership through curiosity and reasoning. He often envisions a future so clearly in his mind that it feels inevitable. This belief allows him to manifest that future despite the difficulties that might occur along the way. Rather than making sudden, large scale changes or issuing brand new mission statements, he builds the foundation for big shifts over many years. He reasons through every step and shares that logic with everyone around him constantly.

I have already made up my mind, but I will take every possible opportunity, external information, new insights, new discoveries, and new engineering revelations. I will use it to shape everybody else's belief system. I am doing that literally every single day with my board, with my management team, and with my employees.

This process of shaping beliefs ensures that when a major decision is finally announced, it feels completely obvious. Jensen describes this as leading from behind. By the time he declares a major pivot, such as going all in on deep learning, he wants his employees to ask what took him so long. This strategy extends beyond the company to partners and the entire industry. Since Nvidia functions as a computing platform that integrates into other companies' products, Jensen must convince the world of a future before his products are even ready.

I am also shaping the belief system of my partners and the industry, and I am using that to shape the belief system of my own employees. By the time that I announce something, I have been talking about the stepping stones for two and a half years.

Jensen describes how AI progress is driven by four distinct scaling laws: pre-training, post-training, test-time, and agentic scaling. While there was early concern that a lack of high-quality human data would limit intelligence, synthetic data has solved this bottleneck. Most information humans share is already synthetic because it is created and modified by people rather than found in nature. AI can now use ground truth to generate its own data, meaning compute power is now the primary limit on training rather than the availability of data.

Thinking is way harder than reading. Pre-training is just memorization and generalization. You are reading. Reading versus thinking, reasoning, solving problems, taking unexplored experiences, and breaking it down into solvable pieces.

Test-time scaling focuses on the inference phase, which Jensen views as active thinking rather than simple retrieval. This process involves reasoning, planning, and searching for solutions, making it intensely compute-heavy. Beyond this lies agentic scaling, where AI systems spawn teams of sub-agents to conduct research and use tools. This mimics the way a company scales by hiring more staff. These agents create new experiences and data that feed back into the original models, creating a continuous cycle of improvement.

A significant hurdle is the speed of innovation. AI model architectures change every few months, but hardware takes years to build. Jensen addresses this by conducting internal research and working with almost every AI company in the industry. This allows Nvidia to anticipate architectural shifts years before they happen. Success requires predicting where the innovation will lead so the hardware is ready when the software catches up.

Success in the rapidly evolving AI landscape requires an architecture that balances incredible acceleration with the flexibility to adapt to changing algorithms. CUDA has remained resilient because it bridges the gap between specialization and generalization. This adaptability allowed for a swift transition from processing mixture of experts models to supporting massive parameter models on a single computing domain. While previous systems like Grace Blackwell focused on large language model inference, the newer Vera Rubin racks are designed specifically for agents that interact with tools and storage accelerators.

The last one was designed to run large language models and inference, and this one is to run agents, and agents bang on tools.

Anticipating the future of AI involves reasoning about the practical needs of a digital worker. Rather than waiting for an AI to become universally smart, it is more effective to give it the ability to conduct research, access ground truth files, and use existing tools. A humanoid robot in a house is more likely to use a microwave by reading the manual than to beam microwaves from its fingers. In the same way, AI will not destroy software but will instead become a sophisticated user of it. Open source projects like OpenClaude represent a pivotal moment for agentic systems, similar to the impact ChatGPT had on generative AI.

As these agentic systems gain the ability to access sensitive data and execute code, security becomes a primary concern. Jensen highlights a safety framework where systems are restricted to two out of three critical capabilities: accessing sensitive information, executing code, or communicating externally. By ensuring an agent never has all three simultaneously, enterprises can maintain control while leveraging the power of AI agents.

Agentic systems can access sensitive information, it can execute code, and it can communicate externally. We could keep things safe if we gave you two out of those three capabilities at any time, but not all three.

While Moore's Law would have improved computing by about 100 times over the last decade, NVIDIA scaled computing by a million times through extreme co-design. This approach focuses on improving energy efficiency every year. Jensen explains that the goal is to drive token costs down as fast as possible. Even as computer prices rise, the effectiveness of generating tokens is increasing much faster. This results in token costs dropping by an order of magnitude every single year.

Power remains a significant hurdle for the future of AI. Jensen considers various solutions, including small modular nuclear power plants. To manage supply chain bottlenecks, Jensen spends significant time informing and inspiring CEOs across the IT and infrastructure industries. He acts as a guide for both upstream suppliers and downstream partners.

I spent a lot of time informing all the CEOs that I work with. What are the dynamics that's going to cause the growth to continue or even accelerate? I am also describing where are we going to go next so that they could use all of this information and all of the dynamics that are here to inform how they want to invest.

Years ago, Jensen had to convince memory manufacturers that High Bandwidth Memory, or HBM, would become a mainstream requirement for data centers. At the time, it was only used in niche supercomputers. He also encouraged the use of low-power mobile memory for data center applications. Today, these components are critical to the industry. The scale of the engineering is immense. A single server rack contains up to 1.5 million components sourced from 200 different suppliers. Jensen manages these intricate relationships to ensure the supply chain remains stable.

NVIDIA changed its architecture from the DGX1 to NVLink 72 rack-scale computing. This shift moved supercomputer integration from the data center directly into the supply chain. Because these systems are so dense, they are now shipped in racks weighing two to three tons. This change also moves the power requirements. If the goal is to produce 50 gigawatts of supercomputing capacity, the supply chain needs a gigawatt of power every week just to test the machines before they ship.

NVLink 72 literally builds supercomputers in the supply chain and ships them 2, 3 tons at a time per rack. It used to be they used to come in parts and we used to assemble them inside the data center. But that's impossible now because NVLink 72 is so dense.

Jensen convinces partners to make massive capital investments by explaining the logic behind these shifts. He focuses on the coming inflection point for the AI inference market. By reasoning from first principles and drawing pictures, he builds a shared view of the future that gives partners the confidence to invest.

Scaling a global supply chain requires deep trust and clear communication with partners. While concerns exist regarding potential bottlenecks in tooling or advanced packaging, Jensen maintains confidence in the ability of partners to meet demand. The key is establishing a mutual understanding of requirements and a shared commitment to the plan.

I told them what I needed, they understood what I need, they told me what they are going to do and I believe what they are going to do.

When partners are aligned on goals and expectations, the focus shifts from worry to execution. This level of transparency and mutual belief allows a company to scale rapidly even during periods of accelerating growth.

The power grid is currently designed to handle worst-case conditions, such as extreme weather days in winter or summer. Because of this, the grid often runs at only about 60% of its peak capacity. This means there is a massive amount of excess power sitting idle 99% of the time. Jensen suggests that data centers could solve this by using that excess power instead of demanding a constant, perfect supply.

99% of the time our power grid has excess power and they are just sitting idle, but they have to be there sitting idle because just in case, when the time comes, hospitals have to be powered and infrastructure has to be powered.

To make this work, Jensen outlines a three-way solution. First, CEOs and end customers must realize that demanding 100% uptime puts unnecessary pressure on the grid. If they accept slightly slower responses during rare peak moments, the whole system becomes more efficient. Second, data centers must be engineered to gracefully degrade. They should be able to shift workloads or reduce power consumption without losing data. Finally, utility companies need to offer different tiers of power delivery. Instead of making customers wait years for grid upgrades, they could provide immediate access to excess power at a different price and guarantee level.

If the utility of the grid tells us, listen, we are going to have to back you down to about 80%. We are going to say that is no problem at all. We are just going to move our workload around. We are going to make sure that data is never lost.

Elon Musk succeeds by being a minimalist systems thinker who questions every requirement. He asks if a step is necessary and if it must take as long as it does. He often goes directly to the point of action to see problems for himself. This creates a sense of urgency that makes his projects a priority for everyone involved. Being on the ground allows an engineer to build an intuition for where inefficiencies exist at both a detailed and a broad scale.

Jensen uses a method called speed of light thinking. This means comparing every task to the physical limits of physics. Instead of looking for small improvements, he prefers to strip a process back to zero. He asks how long a task would take if built from scratch today. This approach often reveals that a process taking many weeks could theoretically be done in days.

I don't love the other methods, which is continuous improvement. I'd rather strip it all back to zero. First of all, explain to me why it is 74 days in the first place. Let's think about what is possible today. And if I were to build it completely from scratch, how long would it take?

Modern systems like the Vera Rubin POD are incredibly complex. They contain millions of components and trillions of transistors. Jensen aims for these designs to be as complex as necessary but as simple as possible. Engineering at this scale requires collaboration between world class teams at companies like NVIDIA, TSMC, and ASML. These systems are the most complex computers the world has ever made.

China has transformed into a global technology powerhouse by leveraging a massive talent pool. About half of the world's AI researchers are Chinese, and many continue to work within the country. This tech industry matured during the mobile cloud era, making the workforce exceptionally comfortable with modern software development. Unlike a single monolithic economy, China functions as a collection of competing provinces and cities. Mayors compete with one another, which fuels the birth of countless companies in sectors like electric vehicles and artificial intelligence. This internal rivalry ensures that the companies that eventually emerge as leaders are incredibly resilient and capable.

China is not one giant economic country. It's got many provinces and cities with mayors all competing with each other. That's the reason why there's so many EV companies. That's the reason why there's so many AI companies. As a result, they have insane competition internally, and what remains is an incredible company.

Cultural factors also play a significant role in this rapid innovation. Jensen notes that the schoolmate bond is a powerful force in Chinese society, where a classmate is often seen as a brother for life. This creates a natural environment for knowledge sharing. Because engineers have friends and family spread across different firms, keeping technology hidden is difficult. Consequently, the industry embraces open source to amplify and accelerate progress. While many Western leaders come from legal backgrounds, many Chinese leaders are engineers who focus on building and growth to move the country away from poverty.

The schoolmate concept is a one schoolmate, your brother for life. And so they share knowledge very, very quickly. And so there's no sense keeping technology hidden. You might as well put it on open source. And so the open source community then amplifies, accelerates the innovation process.

Nvidia pursues a strategy of extreme co-design by conducting basic research in model architecture. This research provides essential visibility into the types of computing systems required for future AI models. For example, Nemotron 3 is not a simple transformer model. It combines transformers with State Space Models. By experimenting with different domains and architectures like conditional GANs, the company can build hardware specifically optimized for how AI is evolving.

One of the things that I love about Nemotron 3 is it is not just a pure transformer model. It is transformer and SSMs. The fact that we are doing basic research in model architecture and in different domains gives us visibility into what kind of computing systems would do a good job for future models. It is part of our extreme co-design strategy.

Jensen believes that open source is fundamentally necessary for the AI revolution to reach every industry and country. While proprietary models function as excellent products, open source allows researchers and students to innovate. This diffusion of technology ensures that AI is not restricted to a few companies. Nvidia shares not only the models and weights but also the data and the process used to create them.

We want world class models as products and they should be proprietary. On the other hand, we also want AI to diffuse into every industry and every country, every researcher, every student. If everything is proprietary, it is hard to do research and it is hard to innovate.

The vision for AI extends far beyond language models. Future systems will likely act as sub-agents trained on biology, chemistry, and the laws of physics. Nvidia aims to push the frontier of these physical AIs to support other industries. The goal is to provide the foundational tools so that car companies can build better vehicles and pharmaceutical companies can discover new drugs without Nvidia needing to enter those specific markets themselves.

A common misunderstanding about TSMC is that its success relies solely on having the best transistors or packaging technology. While its technical capabilities are special, its true brilliance lies in the orchestration of global demand. The company manages the shifting needs of hundreds of global partners while maintaining high throughput and excellent manufacturing yields. They ensure wafers arrive exactly when promised so that their customers can run their businesses reliably.

The deepest misunderstanding about TSMC is that their technology is all they have. Their ability to orchestrate the demands, the dynamic demands of hundreds of companies in the world as they're moving up, shifting out, increasing, decreasing, pushing out, pulling in... somehow they're running a factory with high throughput, high yield, really great costs, excellent customer service.

The culture at TSMC is unique because it balances two traits that rarely exist together. They remain at the bleeding edge of technology while staying intensely focused on customer service. Most companies excel at one but struggle with the other. Above all, the partnership is built on an intangible sense of trust. Jensen notes that despite thirty years of partnership and hundreds of billions of dollars in business, NVIDIA and TSMC operate without a formal contract.

Three decades, I don't know how many tens, hundreds of billions of dollars of business we've done through them and we don't have a contract. That's pretty great.

Jensen recalls being offered the chief executive position at TSMC by Morris Chang in 2013. Although deeply honored, he declined because he felt a sole responsibility to realize his vision for Nvidia. He saw the potential impact the company could have and felt he needed to stay to make it happen. Today, he views both companies as among the most consequential in history.

The work that I'm doing here is really important. I've seen in my mind's eye what Nvidia was going to be and what the impact that we could have. It was really important work and it's my responsibility, my sole responsibility to make this happen.

Nvidia's primary competitive advantage is its massive installed base, centered on the CUDA computing platform. This advantage is not just about technical specs. It is about the trust and dedication of millions of developers. These developers choose CUDA because it provides immediate reach across every cloud and industry. This creates a cycle where developers trust Nvidia to maintain and improve the software for the long term. This ecosystem is vertically integrated but horizontally distributed across providers like Amazon, Google, and Microsoft.

The business has also shifted its focus from individual chips to entire AI factories. The basic unit of computing used to be a single GPU. It then evolved into a cluster and has now become an entire gigawatt-scale facility. This shift requires thinking about massive infrastructure, including power generation, cooling systems, and complex networking. Jensen now views his work through the lens of these massive pods rather than just the underlying chips.

My mental model is this giant gigawatt thing that has power generation and is connected to the grid. It's got cooling systems and networking of incredible monstrosity things. 10,000 people are in there trying to install it. Hundreds of networking engineers in there, thousands of engineers behind it trying to power it up.

Lex explores the idea of using space for AI compute to solve energy issues. Jensen notes that NVIDIA GPUs are already in space. They are used in satellites that perform high-resolution, centimeter-scale imaging. These satellites generate petabytes of data as they sweep the Earth. Beaming all of that data back to Earth is not practical. AI must be done at the edge to keep only the important information.

You don't want to beam that back down to Earth. It's just petabytes and petabytes of data. You ought to just do AI right there at the edge. Throw away everything you don't need, you've seen before, didn't change, and then just keep the stuff that you need.

Space engineering presents unique challenges. Cooling is difficult because there is no air for conduction or convection. Systems must rely entirely on radiation. Space does offer advantages like constant solar power at the poles, but the environment is harsh. Jensen has engineers working on problems like radiation resistance and hardware redundancy. Software must be designed to degrade gracefully. Instead of breaking completely, a computer in space should simply get slower. While space is a future opportunity, Jensen believes there is still a lot of low-hanging fruit on Earth. He focuses on eliminating waste and utilizing idle power before moving entirely into space.

Make it so that the computer never breaks, it just gets slower.

Nvidia's growth is inevitable because the fundamental nature of computing has changed. Computers used to be retrieval systems. They were like warehouses where people stored pre-recorded files and used search tools to find them. This model focused on storage. Now, the world is moving toward generative computing. These new systems are contextually aware and must generate information in real time.

We went from a retrieval based computing system to a generative based computing system. We are going to need a lot more processing in this new world than in the old world. We need a lot of storage in the old world, we need a lot of computation in this new world.

This transition changes the role of the computer from a storage unit to a factory. A warehouse stores items, but a factory produces products that generate revenue. Jensen believes that intelligence is now a scalable product. Just as there are different tiers of physical goods, there will be different levels of tokens. These tokens are units of intelligence that range from free to premium. Jensen notes that people will soon be willing to pay one thousand dollars for a million tokens for specialized high intelligence products. Computation is no longer just an expense. It is a direct driver of productivity and profit.

Its purpose in the world changed. It is no longer a computer, it is a factory. A factory is used for generation of revenues. We are now seeing not only is this factory generating products that people want to consume, we are seeing that the commodities are so valuable that the tokens are starting to segment.

Jensen expects global GDP to accelerate as these AI factories create new drugs and services. In this new economy, the percentage of GDP used for computation could be a hundred times larger than it was in the past. As computers become production units rather than storage units, the scale of the industry will grow significantly.

Jensen believes there are no physical limits preventing Nvidia from reaching trillions in revenue. The company scales its operations through a massive ecosystem of 200 partners. While some critics argue that certain growth targets are theoretically impossible for a semiconductor company, these views often lack first-principles thinking. Jensen looks at the size of the opportunity the company can create rather than just fighting for existing market share.

I still remember the first time we crossed a billion dollars. I was reminded of a CEO who told me it is theoretically impossible for a fabless semiconductor company to exceed a billion dollars. Of course, that is illogical. It is hard for people to imagine how large we could be because there is nobody I could take share from. The challenge for the world is the imagination of the future.

The arrival of advanced AI agents represents the iPhone moment for the industry. These agents are the fastest-growing applications in history. In the future, people will interact with AI by simply talking to their devices, which is a highly efficient way to complete tasks. These AI assistants will likely contact their users frequently because they work so fast. They will report completed work and ask for the next assignment immediately.

Jensen feels the weight of Nvidia's success because of its importance to national security, the economy, and individual investors. He manages this immense pressure through a process of decomposition. He breaks every complex problem down into manageable tasks. Once he identifies a risk, he ensures it is communicated to someone who can take action.

I reason about the circumstance. What has changed, what is hard, and what am I going to do about it? I break it down, decompose the problem, and the decomposition of these circumstances turns it into manageable things that I can do. If you didn't do it, and you didn't get anybody else to do it, then stop crying about it.

By making a list of everything that needs to be done, Jensen can remain calm and sleep at night. He is tough on himself but finds peace by knowing he has done everything possible to protect his partners and the industry. After a problem is broken down and assigned, there is nothing left to do but execute.

Building a massive company like Nvidia involves intense psychological low points. Jensen manages these moments by breaking problems into smaller pieces and sharing the burden with his team. He believes that worrying alone is less effective than decomposing a problem and inspiring others to help solve it. A key part of this process is what he calls systematic forgetting. Just as AI models must forget certain data to learn effectively, people must let go of past setbacks and embarrassments to focus on the future.

One of the most important attributes of AI learning is systematic forgetting. You need to know when to forget some things. You can't memorize everything. You can't keep everything, and you don't want to carry everything.

Jensen describes the mind of a child as a superpower. When faced with a project that might cost billions or take years, he often asks himself how hard it could be. This simple mindset prevents him from over-simulating the pain and humiliations that might occur. If he knew exactly how difficult the journey would be at the start, he might never begin. Instead, he focuses on the light of the future and assumes his goals are achievable as long as his core assumptions remain true.

You don't want to over simulate everything. All the setbacks and the trials and the disappointments, you don't want to simulate all that in advance. You don't want to know that. You want to go into a new experience thinking it's going to be perfect, it's going to be great, it's going to be incredibly fun.

Resilience stems from a combination of starting with a fresh mind, forgetting failures, and staying true to your beliefs. Jensen also maintains humility by observing and learning from others. He treats every observation as an opportunity to simulate how someone else handled a situation well, which allows for constant learning and adaptation.

Success and wealth do not necessarily make it harder to stay humble or listen to others. In fact, performing work publicly provides constant opportunities to be humbled because when a mistake is made, everyone sees it. This visibility encourages a careful and circumspect approach to sharing ideas externally, as those words impact others. Internally, Jensen believes leadership involves constant reasoning in front of people rather than just delivering conclusions.

I am constantly reasoning in front of people and even when I am talking to you, you can see me reasoning through things. I show you the steps that I got there and then you could decide whether you believe what I said in the end.

Jensen approaches leadership by walking employees through his thought process. This transparent method allows others to intercept and disagree with specific steps in the logic rather than just the final outcome. It creates a collective path searching method where the team can steer the direction together. Lex observes that maintaining a tolerance for embarrassment is essential for this type of growth. Admitting when an idea is wrong, even after declaring it publicly, allows for personal and professional development. This grounded perspective often stems from a history of hard work, including early jobs like cleaning toilets, which helps maintain a sense of humility regardless of current status.

GeForce is the primary marketing strategy for NVIDIA. Many people first encounter the brand while playing games like Fortnite or Call of Duty during their teenage years. As they progress to college and enter the workforce, they transition to using professional tools like CUDA, Blender, and Autodesk. This creates a long-term connection with the hardware that begins with entertainment and grows into industry-standard computing.

There has been some discussion regarding DLSS5. Some gamers worry that it might create generic visuals often called AI slop. Jensen understands this concern but explains that DLSS5 is designed to be different. It is conditioned by 3D geometry and ground truth data provided by the artists. The system enhances the visuals without altering the underlying structure or the original artistic intent.

DLSS is integrated with the artist. So it is about giving the artist the tool of AI, the tool of generative AI. They could decide not to use it.

The goal is to provide artists with more expressive capabilities. For example, artists could eventually prompt the system to apply specific styles, like a toon shader, while keeping the artistry consistent. Jensen views the current sensitivity to AI-generated content as a positive development. It helps people realize that human beauty often lies in imperfections. AI is simply another tool in the artist's kit, much like the shaders used to make digital skin look more realistic.

Doom stands as perhaps the most influential game ever made because it transformed the identity of the personal computer. Before its release, the PC was primarily seen as an office automation tool. Doom shifted that cultural perception, turning the machine into a gaming device for families and individuals. While flight simulation games existed earlier, they lacked the widespread popularity required to pivot the entire industry toward 3D gaming. From a purely technical perspective, Virtual Fighter also holds significant importance in the evolution of game technology.

Doom was really impactful. It turned the PC from an office automation tool into a personal computer for families and gamers.

Modern gaming continues to push boundaries with titles like Cyberpunk 2077, which utilizes fully ray-traced graphics. However, the longevity of a game often comes from its community. Jensen is a fan of Skyrim because of its extensive modding scene. Modding allows players to re-experience a familiar world in entirely new ways, effectively making an old game feel fresh over and over again. To support this, Nvidia created a tool called RTX Remix that lets the community inject modern technology into classic titles. While beautiful graphics enhance immersion and transport players to another world, the core of a great game still relies on story and character development.

Jensen believes that AGI has already arrived in certain contexts. An AI might not be able to build a lasting empire like NVIDIA yet. However, it can create a viral web service or a digital influencer that generates significant revenue. These types of businesses can scale quickly and achieve high valuations before fading away.

I think it's now. I think we've achieved AGI. It is not out of the question that an AI was able to create a web service, some interesting little app that all of a sudden a few billion people used for 50 cents, and then it went out of business again shortly after.

Building a complex, long-term company still requires human innovation and management. Jensen addresses concerns about AI replacing workers by explaining that the tools of a job change, but the core purpose of the role often stays the same. He reflects on his own long career to illustrate this point.

The purpose of your job and the tasks and tools that you use to do your job are related, not the same. I've been doing my job for 33 years. I'm the longest running tech CEO in the world. And the tools that I've used to do my job have changed continuously.

Many people predicted that radiologists would lose their jobs once AI computer vision reached superhuman levels. Even though AI surpassed human vision years ago and is now in every radiology platform, the number of radiologists has actually grown. This happened because the purpose of a radiologist is to diagnose disease and help patients, not just scan images. When tools make scanning faster, hospitals can see more patients and perform more diagnoses. This increased efficiency creates a shortage of radiologists rather than a surplus.

The purpose of a radiologist is to diagnose disease and help patients and doctors diagnose disease. And because we are able to study scans so much faster now, you could study more scans, you could diagnose better, you can see people more.

Jensen believes software engineering will follow a similar path. The number of engineers at Nvidia is expected to grow because their true purpose is solving problems, not just writing lines of code. Coding is becoming a process of specifying specifications. This shift could expand the number of people who can code from 30 million to 1 billion. In this future, a carpenter with AI becomes an architect and an accountant becomes a financial advisor. Their artistry is elevated because the tools handle the technical execution.

I wanted my software engineers to solve problems. I didn't care how many lines of code they wrote. Solving problems, working as a team, diagnosing problems, evaluating the result, looking for new problems to solve, innovation, connecting dots, none of that stuff is going to go away.

The artistry of the future lies in knowing how much to specify. Jensen explains that he often under-specifies his directions to employees on purpose. This allows thousands of talented people to make the outcome even better than he could have imagined. Coding will exist on a spectrum. You might be very prescriptive for a specific result or more exploratory to push the boundaries of creativity with an AI agent.

Many people feel significant anxiety about their jobs as AI and automation continue to advance. This is especially true in the white collar sector where the future feels uncertain. Lex acknowledges that these transformative times often bring real suffering to individuals and families who lose their positions. While new technology usually creates more opportunities and makes work more productive, the transition period can be painful.

Jensen approaches this uncertainty by breaking down problems into things he can control and things he cannot. His primary advice for anyone worried about their career is to become an expert in using AI tools. Whether someone is a college graduate, an accountant, a lawyer, or even a carpenter, they should explore how AI can elevate their work. If Jensen were hiring today, he would always choose the candidate who understands how to leverage AI over the one who does not.

If your job is the task, then you are very highly going to be disrupted. If your job's purpose includes using certain tasks, then it is vital that you go learn how to use AI to automate those tasks.

The distinction between a job and a task is crucial. If a role is defined strictly by a single repetitive task, it is at high risk of disruption. However, if that task is just a means to a larger purpose, AI becomes a powerful tool to automate the boring parts and leave the creative responsibility to the human. AI also serves as an effective life coach for practical problems. Lex notes that instead of feeling stuck, a person can talk to a chatbot to break down their anxieties into a point by point plan. This technology provides a form of hand holding that removes the initial friction of being a beginner.

Jensen observes that while AI can recognize and understand human emotions like anxiety or nervousness, the chips themselves do not feel them. Two computers given the same context might produce different statistical outcomes, but it is not because they had a subjective experience. This subjective feeling of falling in love, fearing death, or experiencing the pain of loss remains a unique human mystery. Lex notes that the scale of AI development has been miraculous, but Jensen emphasizes the need to separate the functional loop of intelligence from the broader concept of humanity.

Intelligence is a functional thing. Humanity is not specified functionally. It is a much, much bigger word. And our life experience, our tolerance for pain, our determination, those are different words in intelligence.

Intelligence is becoming a commodity. Jensen points out that he is often surrounded by people more intelligent or educated than he is, yet he successfully orchestrates their work. He encourages people not to feel anxious about the democratization of intelligence. Instead, society should elevate values like character, compassion, and generosity. These human traits are what he considers to be true superhuman powers. AI should be viewed as an incredible tool that makes humans more powerful rather than something that replaces the human experience.

When discussing the future of his work at NVIDIA, Jensen approaches the idea of succession planning differently than most. He does not focus on a single person taking over one day. Instead, he believes in passing on knowledge and reasoning to his team continuously. Every meeting is an opportunity to share insights and skills as fast as possible. This ensures that the collective capacity of the company grows every day through a constant flow of information.

The most important thing you should do today if you care about the future of your company post you is to pass on knowledge, information, insight, skills, experience as often and continuously as you can.

Jensen views his role as a constant teacher and empowerer. He focuses on passing knowledge to others the moment he acquires it. By pointing people toward new discoveries, he elevates the capability of everyone around him. His dedication to this mission is personal. He expresses a desire to stay involved in the work for the rest of his life because of the fulfillment it brings.

Before I even finish learning all of it myself, I've already pointing it to somebody else. Get on this. This is so cool. You're going to want to learn this. And so I'm constantly passing knowledge, empowering people, elevating the capability of everybody around me.

The continuous rate of innovation at NVIDIA is a point of inspiration. Lex observes that the company's work represents a celebration of humanity and great engineering. It is not just about technology but about the spirit of builders who push civilization forward. This culture of excellence makes the engineering process itself something special to witness.

Jensen holds a deep confidence in the kindness and compassion of the human spirit. He operates on the starting assumption that people generally want to do good and help others. While this perspective sometimes leads to being taken advantage of, his experiences consistently prove that the human capacity for good often exceeds expectations. This fundamental belief in humanity fuels an optimistic outlook on what can be achieved in the coming decades.

I've always had great confidence in the kindness, the generosity, the compassion, the human capacity. I start with always that people want to do good, people want to help others. And vastly I am proven right, constantly proven right and often exceeds my expectations.

The current era provides reasons to be romantic about the future. Within a single lifetime, it is reasonable to expect the end of major diseases and a drastic reduction in pollution. Jensen envisions a future where space exploration evolves through the use of humanoid robots. He imagines sending a humanoid on a spaceship and eventually transmitting his own consciousness, captured from a lifetime of digital data and AI training, at the speed of light to catch up with that robot.

Scientific milestones are also rapidly approaching. The mystery of the biological machine and the complexities of neurobiology are likely to be understood within the next five years. Cracking open theoretical physics and explaining consciousness are within reach, turning long standing scientific questions into solvable problems.