It's NOT Intel You Should Be Watching. The REAL PC Threat is Nvidia.

Kind: captions Language: en The computer you've used for years is disappearing. That screen, the icons, the windows, the menus you click, they're all being forced into the background. They're being replaced by a $70 billion transistor black box that makes more and more decisions for you and charges an energy tax every time it does. Nvidia isn't just releasing a new chip. They're trying to do to the CPU what they already did to the GPU. And Microsoft just handed them the keys. A laptop sitting on a store shelf isn't really a computer in the traditional sense anymore. Sure, it looks like one, it turns on like one, but the architecture that defined personal computing for decades is starting to disappear. At the center is a chip Nvidia calls RTX Spark. For years, computers were built from separate parts. A CPU for logic, a GPU for graphics, and increasingly dedicated hardware for AI. Data constantly moved between them. Nvidia reduced all of that onto a single piece of silicon. The result behaves less like a collection of components and more like a single computing organism. Everything shares the same memory. Everything works from the same pool of data. Nvidia says it can perform a thousand trillion calculations every second. >> [music] >> That number itself doesn't even matter. It's all about the results. A thin laptop with all-day battery life can now run an AI model with 120 billion separate settings inside of it. Work that once filled a room now fits in a backpack. Nvidia claims it'll turn the PC from a tool to a teammate. And they aren't doing it alone. Microsoft willingly supplied the missing piece, the software that lets an AI agent hijack your computer. It watches across your apps. It [music] acts before you ask. The desktop, the icons, and the files and folders you've used your whole life are still there. They're just no longer in charge. For two generations, the CPU sat at the center of personal computing. Now, [music] it's just the supporting cast. And the company whose name was stamped inside almost every computer on Earth didn't lose this battle on your desktop. It [music] lost it somewhere much bigger. For three decades, Intel was at the center of the digital world. Its [music] chips powered everything from office equipment to massive server farms behind nearly every website you've ever opened. If computing had a bleeding heart, Intel was it. Then the floor gave way. Between 2021 and 2025, Intel's share of the data center chip market collapsed from roughly 68% to around 6%. At the same time, Nvidia surged to 86%. >> [music] >> One company practically took over the market. By June 2026, Nvidia was worth in the region of $5 trillion. It was the most [music] valuable company on the planet. No business in history had ever been valued that highly, and no [music] one has came close. When the RTX spark was unveiled, the markets reacted immediately. Nvidia's stock jumped while shares in AMD, Intel, and Qualcomm all fell. For decades, Intel and AMD built chips around a design called x86. Its strength was running instructions one after the other as fast as possible. That is just what traditional software needed. AI plays a completely different game. Instead of a long chain of instructions, it performs enormous amounts of simple math all at once. This is where Nvidia had a head start. Its graphics chips were already built to solve thousands of problems at the same time. That is what rendering a video game requires. When AI became the most important thing in computing, Nvidia didn't have to reinvent itself. The future was already optimized. Intel realized the threat and fought back. It built AI accelerators and launched competing products. It spent years desperately trying to close the gap. The gap kept growing. The company that had spent decades setting the pace for the industry was now struggling to keep up with it. None of that was luck or clever marketing. It was the work that changed everything. Nvidia was already waiting. Jensen Huang called it a reinvention of the computer as significant as the day the phone became the smartphone. But winning the data center wasn't enough. To take over the computer on your desk, Nvidia needed something else, control. Nvidia never had to build the best laptop in the world. It only had to make sure every laptop worth buying ran on its products. It also had to ensure that nobody else owned a single piece of it. So, it built the whole thing itself. Nvidia designed the graphics and the AI hardware. It partnered with MediaTek on the processor. TSMC manufactured the chip. Microsoft shaped Windows around it. Before a single laptop reached the store shelf, everything had already been decided. And that left companies like Dell, HP, Lenovo and Asus in a strange position. They're the companies you actually buy from, but all they got was the same finished module. They design the case, the cooling, the battery, and the ports. >> [music] >> They load Windows onto it, and they ship it. They don't own the processor, the graphics, or the AI software that decides what the machine can do. The brands on the lid are just the people who assembled the box. Nvidia keeps around 75 cents of gross profit on every dollar it sells. >> [music] >> That's almost unheard of for a company that makes physical hardware. The companies assembling the laptops live on a few percent. The rest of the value is gone before the device ever leaves the factory. So, every time more units ship, the imbalance grows. None of it shows up in the price tag, but it decides who holds the power. There's another element to this power shift. Almost every major AI system on Earth is built on Nvidia software platform, CUDA. And once your tools are built on CUDA, switching becomes almost like starting over. Nvidia even tried to expand that control further, attempting to buy ARM, the company behind a rival processor architecture, in a $40 billion deal. Regulators shut the move down. So, instead of owning the rival, Nvidia made its ecosystem the default, [music] and the industry just adjusted around it. It's the exact move the companies pulled before. With graphics cards, it absorbed the value that used to be shared across everyone building them. Now, it's running the same play on the processor, the last major part of the PC it didn't already control. A shift like this only works if there is nowhere left to go or backup platform. Microsoft handled that next part. Your current laptop may already be locked out of the future and you'd have no way of knowing it from the screen. To run Windows new wave of AI features, a computer now needs a dedicated AI chip rated at at least 40 tops. That's the measure of how many AI calculations it can process every second. It also needs 16 GB of memory and 256 GB of storage. Specs that many older machines don't meet. Cross the threshold and the features switch on. Fall short and they just disappear into a cloud or don't run at all. These aren't minor additions. It's things like live translation while you speak, image generation on your device, smarter search options and a memory that can look back through all of your work. None of it works without the hardware Microsoft demands. When the requirements were announced, the chips [music] that met them were arms, not Intels. So, the entire industry began shifting toward arm, the same foundation Nvidia and its partners were already built on. The timing wasn't subtle. Microsoft ended all support for Windows 10 in October 2025. That alone pushed a huge wave of people toward buying something [music] new, whether they wanted to or not. The result was the great refresh, the biggest forced hardware swap the industry had seen in years and the machines that clear the bar most comfortably are the ones built on Nvidia's design. Older programs written for x86 don't run directly on these new chips. [music] Windows translates them on the go through a layer called Prism, so they still open. They just run slower and they never get the full power of the silicon underneath. The old software works as a guest. Of course, RTX Spark sails past Microsoft's [music] benchmark. The baseline is 40 tops and Nvidia says this chip reaches over a [music] thousand. That's more than 20 times what Windows asks for and that's the point. The chip isn't built to barely meet today's features. It's built for [music] what comes after them. Everything below that line is already behind. At that level, the chip can run AI models too large for a normal laptop and keep an assistant running continuously, even when you're not at the keyboard. Once most new software is built for that higher level by default, upgrading stops being optional. These kinds of new computers never fully stop. The AI assistants are partly awake, waiting for something to do. Some systems are even designed to keep working [music] overnight, finishing the long tasks while you sleep. That comes at a cost. How much has been heavily debated. The International Energy Agency and repeated by the Electric Power Research Institute in 2024 put a single AI request at around [music] 2.9 watt hours of electricity, compared to roughly 0.3 for a basic web search. That's about 10 times more. So, the billions of queries each day could work out at nearly 10 terawatt hours of extra demand a year. Other researchers disagree. The research institute Epoch AI, along with OpenAI's Sam Altman, say the real figure is much lower. Newer models, they argued, had cut a typical request back down to about the level of an ordinary search. So, the cost per question is falling fast. But, how often we ask is rising. There's an old idea in economics that when something becomes cheaper and easier to use, people don't use less of it, they use more of it, and the total rises anyway. An assistant that takes no effort to summon gets used constantly, even for the small tasks you'd have never bothered a computer with before. So, the cost isn't sitting in some distant data center that you never think about. It is sitting in your lap. The machine runs hotter. The fan kicks on more often. The battery that used to last all day starts asking for a charger by mid-afternoon. And the effects show up elsewhere. Microsoft's own emissions have risen by nearly 30% since 2020, driven by the data centers powering [music] these systems. The more capable the software becomes, the more electricity it burns in the background every time you use it. It's a problem every company ran into at the same time. A traditional processor is great at following rules. It goes through instructions one at a time in order, making a clear yes or no decision. That was perfect for everything from a spreadsheet to flight booking, where each move waits on the one before it. Modern AI doesn't think in clean steps at all. It works by guessing. [music] It runs staggering amounts of basic arithmetic across billions of values to land on the answer that fits the best. Every word an AI writes sets off another round of that weighing, with billions of numbers checked at once. A single short answer can take trillions of those tiny sums. There's no neat line of logic to follow, just a huge cloud of maybes resolved in an instant. Work like that rewards a completely different kind of processor. You don't want a single processor solving problems one by one. You want thousands of small workers doing the same simple calculation at the same time. A traditional processor doesn't disappear. It just becomes the coordinator, handing the work off to a much larger AI section, and handling the small tasks that [music] still need strict order and precision. Running thousands of those calculations in parallel instead of a few in a line is what makes it all fit on a laptop instead of in a warehouse. And once the machine does the reasoning, the person who used to do it [music] starts to change as well. For decades, using a computer meant knowing where things were. You opened a program, you dug through folders, you clicked menus, and you set things up for yourself. The assistant on these new machines is built to make all of that disappear. Given enough time and it remembers yesterday's work, picks the right tool to complete the task, and hands back a finished result. By the time you even sit down, the assistant may have sorted your messages, drafted the easy replies, and lined up what it thinks matters next. The desktop full of windows [music] you once had to navigate becomes something handled in the background. You don't lose control entirely. You can still reject a step. What disappears is the visible control that defined the old era. Knowing where the files are, knowing what's running, knowing why the machine is doing what it's doing. The chip doing the thinking was always a sealed box that no one could look inside. Now, the way you talk to it is sealed, too. The understanding that people built about how their computers work becomes less and less necessary. [music] And a skill you don't use is a skill you stop remembering. A laptop bought in 2024 or early 2025 is already a relic. Open it up and you can still see how it runs. You can change the settings by hand. You can edit the files that control how it behaves. It belonged to you and it answered to you. All the work an RTX Spark chip does happens inside a sealed box. [music] Its entire reasoning is built on a series of odds, not fixed rules. The software is owned by the duopoly of Nvidia and Microsoft. You give it a goal and the assistant [music] acts. The desktop is still there, just for older programs. It's no longer where the computer actually runs. There's no easy way back to the old model. Software gets written for the hardware that runs it the best. Users follow the software. Older machines fall behind and become redundant. The era of the personal computer as a general purpose tool owned and operated by its user [music] is coming to an end. What replaces it is a system where intelligence and access depends on what your machine can run. And no one outside Nvidia can fully see how it produces what it produces. [music] The real divide isn't technical. It's access. Who gets to think with it and who gets left behind. Nvidia and Microsoft are starting to corner the entire market and that affects your ability to afford what goes inside your computer. Watch real reason RAM prices are skyrocketing and getting worse. Or click on this video.