Nvidia is having quite a week.

The tech company, which specializes in making chips used for AI, added more than $277 billion in market value on Thursday, beating the record for single-day market gains previously held by Meta. The surge followed company earnings on Wednesday, in which Nvidia beat analyst expectations and reported revenues of around $22.1 billion last quarter. It is now the third most valuable company in the U.S., behind Apple and Microsoft.

“Accelerated computing and generative AI have hit the tipping point,” Nvidia founder and CEO Jensen Huang wrote this week. “Demand is surging worldwide across companies, industries, and nations.”

But the chipmaker poised to become the most important company in the world has humble beginnings—it was born in a Denny’s restaurant—and its path to success has not been linear. Here are some key leadership lessons from Nvidia and Huang, who embraces fear as a motivator, collaborates with his top managers, and reminds his workers that the company is never more than a month away from shutting down.

Stick to your guns

In navigating the uncertainty of the tech industry, Huang bet on a specific kind of technology—GPUs.

That stands for graphics processing unit, which was previously considered a niche technology and mostly used by gamers. Now, it’s critical to AI, with one analyst likening the AI boom to a war, and calling Nvidia the “only arms dealer out there.”

But Huang bet on GPUs years before it was clear to everyone else that it would pay off, and stuck to that decision through bad times and good. “What he did was recognize that even though there was boom and bust around the GPU, that eventually if there were enough use cases of it, they would offset each other,” Gene Munster, managing partner at Deepwater Asset Management, told Fortune.

“What sets Jensen apart is he’s a wartime CEO,” Munster added. Even when Nvidia is out of favor with investor consensus, and facing near-term setbacks, the company stays “true to [its] view of the future.”

Ignorance is bliss

Huang has openly said that if he knew how hard it would be to create and maintain his company, he probably wouldn’t do it again.

“Building a company and building Nvidia turned out to have been a million times harder than I expected it to be, than any of us expected it to be,” Huang said last year on the Acquired podcast.

He added that ignorance wasn’t just important for him personally—it’s an important quality for all founders.

“I think that’s kind of the superpower of an entrepreneur. They don’t know how hard it is. And they only ask themselves: ‘How hard can it be?’ To this day, I trick my brain into thinking: ‘How hard can it be?’” he said.

Fear is a powerful motivator

Nvidia had some serious setbacks in its early days.

The startup struggled during the dotcom crash in 2002 and the financial crisis in 2008, when demand for its products plummeted.

When everything was going wrong, the company wasn’t shy about making clear to employees how close they were to the brink. In fact, a mantra for Nvidia in its early years was: “Our company is 30 days from going out of business.”

“We have the benefit of building the company from the ground up and having not exaggerated circumstances of nearly going out of business a handful of times,” Huang said last November at Harvard Business Review’s virtual Future of Business conference. “We don’t have to pretend the company is always in peril. The company is always in peril, and we feel it.”

Huang believes it’s critical for leaders to always remember how close they are to failure. “You’re always on the way to going out of business,” he said at Columbia Business School’s Digital Future Initiative last fall. “If you don’t internalize that sensibility, you will go out of business.”

Collaboration is key

Huang couldn’t have amassed his AI and chip empire without collaborating with his top lieutenants.

In forming a system of governance, Nvidia decided to split decision-making among Huang and his fellow founders, Chris Malachowsky and Curtis Priem. Even though Malachowksy and Priem worked under Huang’s command, they each had authority and led their own divisions within the company.

“We would talk or argue over each other’s decisions, but we would default to the final decision of the person who had the expertise in that area,” Priem told the Wall Street Journal in December last year. “It wasn’t ‘agree to disagree.’ The decision terminated any disagreements and became the direction we were going.”

Share your failures

Huang believes that “failure must be shared,” and doesn’t shy away from taking ownership and pointing out errors that led to larger companywide problems.

Back in the early 2000s, Nvidia ran into some trouble with a faulty graphic card that had a loud fan. Instead of firing the product managers, Huang held a meeting where the people involved walked through the mistakes that led to creating a bad product.

Exhibiting one’s failures has become a cultural fixture at Nvidia, but it’s not for everyone.

“You can kind of see right away who is going to last here and who is not,” Dwight Diercks, head of software for Nvidia, told The New Yorker last November. “If someone starts getting defensive, I know they’re not going to make it.”