How Nvidia Owned A.I. | Light Speed or Bust | 1

0:08It's May 2023 and in Manhattan a stock trader hunches over his terminal eyes glued to the stock updates flashing across the screen. NVIDIA is about to release its quarterly results, and this trader hopes to make a killing from it. NVIDIA designs microchips. It started out creating chips that made video games look better. These days, its chips power the AI boom, and its competitors trail far behind. NVIDIA owns nearly 90 % of the market for AI chips. If generative AI is like the California Gold Rush? Well, NVIDIA is the guy selling shovels to miners. You know, the one who makes the real money. The question today isn't whether NVIDIA is doing well, it's how well it's doing.

0:53But this trader needs to move quickly. Today's stock markets are run by computers that trade at lightning speed, often with the help of NVIDIA chips. Against these machines, a blink that delays his trade even by a split second could cost millions. The trader's terminal flashes the breaking news. NVIDIA second quarter revenue estimate 7.2 billion. The trader stares in shock. NVIDIA's projected revenues are four billion dollars higher than expected. NVIDIA hasn't just beaten projections, it crushed them. By the time the trader recovers, the computers have already pounced and made their trades. His chance to make a big profit from the news has vanished in an instant.

1:42The next day, NVIDIA's stock price rises so high that the company's market valuation increases by$184 billion. NVIDIA is now the world's sixth most valuable company, ahead of Visa and Walmart combined. In 13 months' time, it will leapfrog Apple and Microsoft, to become the most valuable company on the planet. And it's all because NVIDIA is the arms dealer of the artificial intelligence wars. It doesn't care if Microsoft, ChatGPT, DeepSeek, Google, Anthropic, or Apple emerge victorious, because no matter who does, NVIDIA wins. But how did a company that set out to make games look prettier, become the biggest winner in the AI revolution?

2:37This message is brought to you by Apple Card. Apple Card members can earn unlimited daily cash back on everyday purchases, wherever they shop. This means you could be earning daily cash on just about anything, like a slice of pizza from your local pizza place or a latte from the corner coffee shop. Apply for Apple Card in the Wallet app to see your credit limit offer in minutes. Subject to credit approval, Apple Card issued by Goldman Sachs Bank USA, Salt Lake City branch. terms and more at applecard.com. As business owners and managers, you use software for your business every day. You use one piece of software to manage your customers, another to manage your employees, another to manage your finances, and the list goes on.

3:19You buy these pieces independently and hope they fit neatly together like a puzzle. And then you find out the hard way that they don't, and you end up with a mess at the heart of your business operations. Does any of this sound familiar? Well, fortunately, Zoho offers a solution to this chaos. It's called Zoho One. Zoho One is a suite of around 50 pre-integrated business applications that fit together beautifully. So instead of dealing with disparate software from multiple vendors with multiple contracts and price points, you deal with one vendor with all the pieces of the business software puzzle neatly put together, offered at a very attractive price.

4:00Now, if this sounds interesting to you, you've got to check out Zoho One at Zoho.one. That's Z-O-H-O dot O-N-E. With Zoho, you're not just licensing apps. You're licensing peace of mind.

4:21From Wondery, I'm David Brown, and this is Business Wars.

4:47With his signature black leather jacket uniform, NVIDIA CEO Jensen Huang has become one of tech's most iconic leaders. He's taken NVIDIA from startup to superstar. And now, riding the wave of AI mania, it's become the world's most valuable company worth trillions of dollars. But Huang's success is an against-the-odds story. NVIDIA entered a crowded market, stumbled badly, and nearly went under. And its ultimate victory came from an expensive bet on a product no one seemed to want. So, how did Huang take NVIDIA to the top? And now that it's there, can he protect its monopoly from rivals? This is Episode 1, Lightspeed or Bust?

5:44It's 1992, and in Mountain View, California, Curtis Priam is in shock. Priam is in his early 30s. He's an engineer at Sun Microsystems, one of Silicon Valley's hottest tech firms. And he's just learned that the team he leads could be dissolved. Sun makes workstations, powerful computers used by technical professionals in fields like industrial design, engineering, and animation. These customers often use their workstations to create and manipulate 3D images. But 3D graphics require lots of computing power. Every pixel, every movement requires complicated calculations that take time to process.

6:27So Priam and his team came up with a solution. Graphics accelerators, specialized microchips that handle the heavy 3D calculations and take the burden off the main processor, the CPU. Priam believes graphics accelerators are the future. Without them, 3D graphics will forever be limited by CPU's processing power. But his bosses at Sun aren't interested in graphics accelerators, so they're shutting down his team. That's when Priam and his colleague, Chris Malachowski, hatch a plan. They'll start their own business making graphics accelerators. Not for corporate computer workstations, but for home PCs.

7:10In fact, they think the time is perfect for a graphics revolution in home computing. People can now easily upgrade their PCs just by plugging in expansion cards. These cards are circuit boards that offer enhancements like better sound chips, modems, or extra memory. And with video games starting to transition from 2D to 3D graphics, Priam and Malachowski believe gamers will soon want to add graphics accelerator cards to their PCs. There's just one problem. Priam and Malachowski know how to engineer chips, but they don't know how to market them, how to raise money, or how to negotiate business deals.

7:50Luckily, they know someone who does.

7:58It's late 1992 in East San Jose, California. Inside a Denny's at a booth with a view of the freeway, Priam tries to ignore the bullet holes in the window. And his breakfast. This is a sketchy neighborhood, and he's not thrilled about the food either. But the man he and Malachowski want to run their startup insisted on meeting here. That man is Jensen Huang. He was born in Taiwan and moved with his family to the U.S. when he was nine. He likes Denny's. He worked there during high school. He credits the job with bringing him out of his shell and teaching him to find satisfaction in his work, no matter the task.

8:38These days, he works for LSI Logic, a company that helps other companies build microchips. He's only 29, but he's already seen as future leadership material. Priam and Malachowski worked with Huang when they were building their graphics accelerator for Sun. He can be prickly, but they like him. and they know he has the drive and business experience their startup needs. But Huang isn't sold on joining them. Look, I know you hate your jobs, but I love mine. Why would I quit? Priam leans forward. Because this is a chance to build something big. How big? I don't know, but big. You play video games, right?

9:22Yes. Then you know games are moving into 3D graphics. It's already happening in arcades. It'll happen for PCs, too. But PCs gotta have graphics acceleration. Well, I agree. But like I told you last time, I'm not interested unless this business can deliver sales of$50 million a year. And I'm not seeing that. Huang pulls out his laptop and shows Priam and Malachowski the spreadsheet he's built. It's a detailed model of the business and offers projections of annual revenues years into the future. But right now, the projections fall short of Huang's magic number. The numbers don't work. I can't tell my wife I'm quitting my job for this unless it has major potential.

10:05She just gave up her career because we couldn't find reliable child care. How can I tell her I want to leave a good job to chase some dream? Priam stares into his coffee. This isn't going to plan. But then Huang notices his disappointment. Look, my main concern right now is the competition. I've identified 35 companies who are already developing graphics accelerators. Does the world really need a 36th? Priam looks up, hopeful again. We can out-engineer them. You know that. Wong shuts his laptop. Yes, but Betamax was better than VHS, and we all know how that ended. Engineering is not enough. So, you're out?

10:50No, but I need to do more research before deciding.

10:59After a few more weeks of research and refining his model, Wong's spreadsheet shows this business can meet his$50 million target. Hey, you see what happened there? Wong's spreadsheet killed a dream, then brought it back to life. That's a lesson in what separates a hunch from a business. Most founders fall in love with their ideas. The smart ones make the math do the talking. It's not that the vision doesn't matter. But a vision you can't model, that's just a daydream. In April 1993, armed with the new spreadsheet, Huang, Priam, and Malachowski found NVIDIA. The name is a shortened version of NVIDIA, the Latin word for envy, which is what they hope their rivals will feel.

11:49But now, they need investors.

11:56It's summer 1993, and NVIDIA's co-founders are pitching their idea to venture capitalist Don Valentine. He's the founder of Sequoia Capital and a legend in the tech world. The man who saw the potential in Atari, in Apple, Electronic Arts, Oracle, and more. But today, he's not seeing much to love in NVIDIA. They've spent most of this pitch showing him a hacked-together prototype hooked up to a virtual reality headset. And Valentine's main takeaway is confusion. He zeroes in on Prem. What are you trying to be? A gaming console maker, a graphics firm, or an audio company? We're all of them. We can integrate it all into one ship.

12:40Well, you have to pick one, otherwise you're going to fail, because you don't know who you are. It's a good point. When you're building something new, it's tempting to be everything to everyone. But customers and investors can smell confusion a mile away. A clear value proposition beats a clever one every time. Pick your lane, own it, and only then think about the next highway. Valentine's colleague Mark Stevens jumps in and tries to move past NVIDIA's unfocused pitch. Let's talk practicalities here. Your plan is to design the chips and license them to a manufacturer who will supply them to graphics card manufacturers who will handle the marketing and distributing to stores.

13:23Is that correct? Quang responds. That's right. We think it's more efficient to sell our chips to third parties who can market them to end consumers. And who will manufacture your chips? We intend to work with SGS Thompson. Valentine and Stephen shake their heads. SGS Thompson is a company riddled with problems. They expected Huang to say TSMC, which is fast becoming the world's leading microchip manufacturer. Valentine sighs. He's heard enough. Well, I think that's all for today. I have to confess, I didn't find your pitch to be very good, and I think you know it already. The NVIDIA team leaves crushed, but Valentine and Steven sense there's something there.

14:12Yes, the pitch was messy, but its founders are talented. And from its previous investments in gaming and graphics, Sequoia knows a potential market exists. But most importantly, Wang came recommended by the founder of LSI Logic. So, after more meetings, Sequoia decides to take a chance and invests$1 million. NVIDIA gets another million from Sutter Hill Ventures. The company opens an office in a strip mall next to a Chinese takeout restaurant and begins developing its first graphics chip, the NV1. Priam wants the NV1 to be revolutionary. Most 3D graphics are created using lots of small triangles and other polygons.

15:01The polygons are then combined to form bigger objects like cars, trees, or people. This is called polygonal rendering. But Priam thinks there's a better way. Quadratic rendering. Instead of using polygons, quadratic rendering builds 3D objects out of curved shapes. As a result, it's better for creating more lifelike images. Priam doesn't stop there. The NV1 can also handle audio, so PC owners won't need to buy a separate sound card. And he finds ways to make the chip use less memory than rival graphics accelerators. Which matters, because in 1993, computer memory costs around$50 per megabyte. That's a lot.

15:45If NVIDIA's chips use less memory, they can undercut their competitors' prices. In May 1995, NVIDIA announces the NV1, and it rapidly generates buzz. Sega hires NVIDIA to help develop the graphics chip for its next games console. It also agrees to adapt its hit arcade game Virtual Fighter for PCs, so it can act as a showcase for the NV1. Nvidia's retail partner Diamond Multimedia orders 250 ,000 NV1 chips and delivers them to stores just in time for Christmas. But by then, cracks are showing in Nvidia's plan. The price of memory has fallen from$50 to$5 per megabyte, wiping out the company's price advantage.

16:34The NV1 also lacks support from most game makers. Few game developers want to spend time and money learning about quadratic rendering or adapting their games to make use of NV1's unique sound features. As a result, games run poorly on the NV1. And unimpressed gamers start returning their cards to stores. Then comes the death blow. Microsoft introduces new standards that graphics accelerators have to follow if they want to be supported by Windows. and these standards require using polygonal rendering instead of the quadratic rendering used by NVIDIA's chip. Following Microsoft's decision, all demand for the NV1 evaporates.

17:21No one wants a graphics accelerator that doesn't work with Windows and makes games look sound and play worse. See, NVIDIA changed innovation so hard, it forgot about compatibility. For small businesses, that's like inventing the perfect frozen pizza. Unless someone's got an oven, it's kind of worthless. The best product seldom wins. The most adoptable one often does. But NVIDIA's learning that lesson way too late. By the end of 1995, the company is facing a quarter of a million returns. It over-engineered its product and misjudged the market. And now, it's going bust.

18:10Closing the books, getting your people paid, and bringing on new hires. Running a small or mid-sized business can be exciting and also a little chaotic. Workday Go makes simplifying your business, well, simple. Imagine all the important aspects of your company, HR, finance, and payroll, all on the AI platform. No more juggling multiple systems. No more worrying about growing too fast. Just the full power of Workday Go, helping small to mid-sized businesses like yours scale and run more smoothly. Think about what that means. Seamless onboarding for new team members. Real-time insights at your fingertips.

18:50And payroll that works perfectly every single time. So you can focus on the big picture and go after your big ambitions. And with Workday, you can activate quickly in as little as 30 to 60 business days. So simplify your business. Go for growth. Go with Workday Go. December can be a busy month for all of us. Finishing projects, planning for next year, maybe even wrapping a few presents while your teams take a well-deserved break. Cyber threat actors don't. Stay protected this holiday season with Manage Engine. Their cybersecurity solutions ensure that your data is secure, even while your office lights are turned off.

19:33With 18 data centers around the world and over 20 years of experience securing global enterprises, Manage Engine can keep your business running safely, no matter how busy or hectic the world gets. This holiday, gift yourself and your business peace of mind. Visit cybersecurity.manageengine.com. That's cybersecurity.mangine.com.

20:19It's 1996, and computer graphics consultant David Kirk has just pulled into a strip mall in Sunnyvale, California. It's his first day as an advisor to NVIDIA, and he's not expecting this gig to last very long. He walks past the Chinese takeout and enters the unit next door, NVIDIA's headquarters. Inside, the vibe is grim. CEO Jensen Huang has just fired more than half of the company's 100 employees. The employees who remain look broken. This is a startup on a deathbed. Kirk's relieved he insisted NVIDIA pay him weekly. Huang greets Kirk and leads him to the back office. Inside is a huge, ugly machine.

21:06It's as big as a refrigerator with exposed circuits and cables that snake out of it like tendrils. This is also NVIDIA's last hope. The machine is a hardware emulator, a computer that can simulate how a microchip design might behave in reality. It cost a million dollars. Almost all the money NVIDIA has left. Huang bought it out of desperation. Usually it takes two years to bring a new graphics accelerator chip to market. But NVIDIA can't wait that long. If it fails to get a new chip out within a year, the company will run out of money. So Huang's looking for shortcuts. Usually, developers make several prototype chips so they can test and fix bugs before putting the chips into production.

21:56Huang plans to skip this stage by testing everything virtually, using the hardware emulator to root out flaws. Kirk thinks this plan is crazy. What Huang's proposing is like trying to hit a hole-in-one in golf. It's possible, but improbable. The emulator can only approximate a chip's real-world performance. If a serious flaw is missed, the whole production run could be ruined. Huang shrugs. If that happens, NVIDIA will go bust. But if he doesn't take the risk, it'll go bust anyhow.

22:38The need to get a new chip made quickly leads Huang to push an approach he calls Speed of Light. The idea is to start every task by imagining the fastest way possible to complete it, even if it's only achievable in theory. Then, work backwards to see how close to this speed you can realistically get. And finally, go all out to meet that timescale. The hardware emulator is just one way to save time. Huang also licenses tech from other companies to shrink development time and harangues employees who don't live up to his demanding expectations. He also scales back NVIDIA's vision. Instead of revolutionizing computer graphics, NVIDIA will focus on making the best possible chips for playing first-person shooter games.

23:27Co-founder Curtis Priam objects to this approach, but Huang and the board overrule him. You see that pivot? That's survival over pride. Huang wasn't abandoning the vision. He was buying time. In a crisis, the smartest play might not be to double down on genius, but just live long enough to use it. Under the new plan, NVIDIA starts developing chips twice as fast as its rivals, which gives them a powerful edge in the 3D graphics arms race. Game developers compete by creating games with ever-flashier visuals, and gamers want the latest tech so they can get the best experience possible. By getting to market quicker, NVIDIA can offer the latest advances sooner.

24:14But to really turn heads, NVIDIA has to outshine the market leader, 3DFX.

24:25It's April 1997, and the Computer Game Developers Conference is about to get underway in Santa Clara, California. In the exhibit hall, NVIDIA's team is setting up their booth when an exhibitor wanders over. His badge says he works for 3DFX, and there's a smirk on his face. Whoa, you guys are still around? Ha, who knew? The NVIDIA team fumes as he walks off chuckling. They know their credibility is shot, but still, they're feeling confident. Nvidia's new chip, the Revo 128, was built in record time, and it's performing great. The only problem is that the demo software they have on display is buggy and prone to crashing.

25:09They just hope it will hold together today. As the conference begins, game developers trickle into the exhibition hall. Most shun Nvidia, but a few take an interest in its tech demo, which showcases the Revo 128's graphics capabilities while also displaying the real-time performance data. And the number of triangles the new NVIDIA chip is moving around is impressive. By lunchtime, word is spreading that NVIDIA's new chip is twice as powerful as the average graphics accelerator. More people start to drop by the stand and they leave as converts. By late afternoon, the buzz has reached 3DFX co-founder and research chief Scott Sellers.

25:55So he heads to Nvidia's stand to see it for himself. Hey there, Scott Sellers, 3DFX. Uh, can I get a demo? The Nvidia rep gulps. He knows Sellers will be looking for ways to convince people that Nvidia's new chip doesn't live up to its hype. The obvious line of attack is to say the demo is fake. The only way to prove it's real is to reboot the demo in front of him, but the buggy software is so unstable, the demo might fail to reload. It's a big gamble, but the rep doesn't have much of a choice. He reboots the PC and smiles awkwardly at Sellers as he waits to see if it will reload. To his relief, the demo restarts without a glitch.

26:42Sellers leans in, watching the performance stats on screen. Five million triangles per second? Wait, that, no, no, no, that can't be right. The NVIDIA rep grins. No, that, that's correct. Seller's face falls. This morning, he thought NVIDIA was just another failed competitor. Now, it's back with a vengeance. Even worse, It'll take his company two years to develop a new chip that can match it. Four months later, NVIDIA's Riva 128 chip reaches stores. By then, the company's just 30 days away from running out of money. But it's not just counting on performance to win gamers over. It's lowering prices, too.