主持人：您应用这项技术的方式具有革命性意义。它使您能够上市 (IPO)，甚至在接下来的四年里将收入增长九倍。但就在取得所有这些成功的同时，您决定根据与一位化学教授的电话交谈来改变 Nvidia 的创新方向。您能告诉我们那个电话的内容以及您如何将听到的内容与您所做的联系起来吗？
黄仁勋（继续）：记住，我们公司核心是开创一种新的计算方式。计算机图形只是第一个应用，但我们一直都知道会有其他应用，例如图像处理、粒子物理、流体等等。我们使处理器更加可编程，以表达更多的算法。然后，我们发明了可编程着色器，使所有形式的成像和计算机图形都可编程。这是一个重大突破。
黄仁勋（继续）：在此基础上，我们还在 2003 年发明了一种名为 CG (C for GPUs) 的东西，它早于 CUDA 大约三年。马克·汉博德 (Mark Hanbold) 写了那本教科书。CG非常酷。我们编写了有关它的教科书，教人们如何使用它，并开发了工具。然后，包括斯坦福大学的学生和 Nvidia 未来工程师在内的几位研究人员开始使用它。
甚至马萨诸塞州总医院的两名医生也将其用于 CT 重建。我飞过去见他们，询问他们正在用它做什么。他们解释了他们如何将其用于研究。然后，一位计算化学家告诉我他使用它来表达他的算法。
主持人：这让我们越来越确信人们可能想要使用这项技术。它强化了我们坚信这种形式的计算可以解决普通计算机无法解决的问题。每次你听到新东西，你都会为惊喜所着迷。这似乎贯穿了您在 Nvidia 的整个领导历程。感觉就像您在技术拐点之前远远地押注一样，当机会最终出现时，您已经准备好抓住它。
黄仁勋（继续）：这看起来确实像一个起跳接球，但你所做的事情基于核心信念。您坚信您可以创造一台可以解决普通处理无法解决问题的计算机。CPU 的能力是有限的，而您可以使用这种新方法解决有趣的问题。
问题总是：这些仅仅是有趣的问题，还是可以成为有趣的市场？因为如果它们不是有趣的市场，那么它们就不可持续。Nvidia 花了大约十年时间投资于这个未来，但市场并不存在。当时只有一个市场：计算机图形学。在 10-15 年的时间里，推动 Nvidia 发展到今天地步的市场根本不存在。
主持人：那么，在您周围的所有人——Nvidia 管理团队、与您一起创造未来的杰出工程师、所有股东、董事会和合作伙伴——的支持下，您将如何继续前进？您带着所有人前进，但目前还没有真正市场的迹象。这项技术能够解决问题，并且由于它可以实现研究论文，这一点很有趣，但您一直在寻找那个市场。
黄仁勋（继续）：尽管如此，在市场存在之前，您仍然需要未来成功的早期指标。我们在公司里有一个说法：
“EOIs FS” - 未来成功早期指标 (Early Indicators of Future Success)。这可以帮助人们，因为我一直在用它给公司带来希望。
主持人:当然。但存在一些重要的问题，这也是公司存在的意义，就是解决这些问题。我们希望可持续发展，因此市场必须在某个时候存在。但是，你想将结果与你做正确事情的证据区分开来，对吧？那么，你如何解决投资于遥远未来的项目并坚持下去的信念？
黄仁勋:你要尽早找到你做正确事情的指标。从一个核心信念开始，除非发生一些改变你想法的事情，否则你将继续相信它，并寻找未来成功的早期迹象。
主持人:英伟达的产品团队使用过哪些早期指标？
黄仁勋:各式各样！在我了解深度学习之前，我看到了一篇关于深度学习的论文。然后，我遇到了一些研究人员，他们需要我们帮助他们为他们的深度学习算法创建特定领域的语言，以便它们可以在我们的处理器上工作。我们创建了 cuDNN，它本质上是神经网络计算的 SQL。我们创建了一种专门用于深度学习的语言，有点像深度学习的 OpenGL。他们需要它来表达他们的数学，他们不懂 CUDA 但懂深度学习，所以我们为他们创建了中间地带。
主持人:即使一开始没有任何经济回报，你还是愿意这样做？
黄仁勋:是的。这是我们公司的一个巨大优势：即使财务回报完全不存在或非常遥远，我们也愿意做一些事情。
我们问自己：这是值得做的工作吗？它是否在某个重要的科学领域取得了进展？我们从工作的意义中找到灵感，而不是从市场的规模中找到灵感。
工作的意义是未来市场的早期指标。没有人需要为此编写商业案例。
唯一的问题是：这项工作重要吗？如果我们不做，它会发生吗？
主持人:你所做的这些选择获得了巨大的回报，但你也不得不引导公司度过一些非常具有挑战性的时期，例如在金融危机期间，由于华尔街不相信你对机器学习的押注，公司市值损失了 80%。在这样的时期，你如何掌舵公司并保持员工的积极性？
黄仁勋:我在那个时候的反应和我今天对这件事的反应是一样的。我的脉搏完全一样。你只需要回到你的信仰。你检查一下自己。什么最重要？你把它们一一检查。家人爱我，对吗？检查。你回到你的核心，然后回到工作岗位，让公司专注于核心。发生了什么变化吗？股价变了，但其他东西变了？没有？然后你继续前进。
主持人:与你的员工交谈时，他们说你尽量避免公开演讲。他们说你的领导风格非常投入。你能告诉我们你为什么特意设计了一个扁平化的组织吗？
黄仁勋:没有任务对我来说是卑微的。我曾经是一名Denny's餐厅的洗碗工！你无法告诉我比这更卑微的任务。如果你给我发一些东西，你想要我的意见，我可以为你服务，并分享我的推理过程。通过推理，我赋予人们力量。我向人们展示如何思考战略、预测和解决问题。这需要花费大量精力，但我从这个过程中获得了回报。首席执行官应该拥有最多的直属下属，因为直接向首席执行官汇报的人需要最少的管理。
我的工作是创造条件，让你能够完成你的毕生工作。这种条件就是赋权，而赋权来自于理解。你需要被告知，而实现这一目标的最佳方式是尽可能减少我们之间的信息层级。这就是我公开推理的原因。这创造了一个高度赋能的组织。英伟达的 30000 名员工是世界上最小的规模化公司，但每个员工都拥有权力并每天做出明智的决策。他们理解背景，因为我对他们敞开，并透明。
这里有个背景知识点：黄仁勋对Denny's餐厅很有感情，他的第一份兼职是在Denny's餐厅做洗碗工，即使是在餐厅中，Busyboy是最卑微的勤杂工，这个工种几乎负责餐厅里最累，最没有技术含量，纯粹体力劳动的工作。而这段餐厅兼职的工作对性格内向且坚韧的黄仁勋是很大的锻炼，他从不掩饰这段经历。黄仁勋有关Denny's餐厅的第二段经历是，黄仁勋曾在 LSI Logic 担任工程师时期遇到了英伟达公司的创始合伙人，而克瑞斯（Chris Malachowsky）和柯蒂斯（Curtis Priem），这两位Nvidia 的联合创始人，之前都在 Sun Microsystems 工作。他们曾经经常在Denny's餐厅讨论创业的事情。
主持人Interviewer:The way you applied this technology turned out to be revolutionary. It got you to the point where you could go public (IPO) and even grow your revenue nine times in the next four years. But in the middle of all this success, you decided to pivot Nvidia's focus on innovation based on a phone call with a chemistry professor. Can you tell us about that phone call and how you connected the dots from what you heard to where you went?
Jensen Huang:Remember, at our core, the company was pioneering a new way of doing computing. Computer graphics was the first application, but we always knew there would be others – image processing, particle physics, fluids, and so on. We made the processor more programmable to express more algorithms. Then, we invented programmable shaders, which made all forms of imaging and computer graphics programmable. That was a major breakthrough.
Jensen Huang:On top of that, we invented something called CG (C for GPUs) in 2003, which predated CUDA by about three years. The same person who wrote the textbook that saved the company, Mark Hanbold, wrote that textbook. Cg was really cool. We wrote textbooks about it, taught people how to use it, and developed tools. Then, several researchers, including students here at Stanford and future engineers at Nvidia, started using it. Even a couple of doctors at Mass General picked it up for CT reconstruction. I flew out to see them and asked what they were doing with it. They explained how they were using it for their research. Then, a computational chemist used it to express his algorithms.
Jensen Huang:This gave us more and more confidence that people might want to use this technology. It reinforced our belief that this form of computing could solve problems that normal computers couldn't. Every time you heard something new, you savored the surprise. This seems to be a theme throughout your leadership at Nvidia. It feels like you make these bets so far in advance of technological inflection points that when the opportunity finally arises, you're there ready to catch it.
Jensen Huang:It does seem like a diving catch, but you do things based on core beliefs. You deeply believe that you could create a computer that solves problems normal processing can't do. There are limits to what a CPU can do, and there are interesting problems you can solve with this new approach. The question is always: are these just interesting problems, or can they also be interesting markets? Because if they're not interesting markets, it's not sustainable. Nvidia went through about a decade where you were investing in this future, and the markets didn't exist. There was only one market at the time: computer graphics. For 10-15 years, the markets that fuel Nvidia today simply didn't exist.
Jensen Huang:So, how do you continue with all the people around you – the Nvidia management team, the amazing engineers creating this future with you, all the shareholders, board of directors, and partners? You're taking everyone with you, but there's no evidence of a real market yet. The fact that the technology can solve problems and research papers are made possible because of it is interesting, but you're always looking for that market.
Jensen Huang:Nonetheless, before a market exists, you still need early indicators of future success. We have a phrase in the company: "EOIs FS" - Early Indicators of Future Success. This helps people because I was using it all the time to give the company hope.
Interviewer:Absolutely. But there were important problems and that's what the company's about, to solve these problems. We want to be sustainable and therefore the markets have to exist at some point. But you want to decouple the result from the evidence that you're doing the right thing, okay? So how do you solve this problem of investing into something that's very far away and having the conviction to stay on the road?
Jensen Huang:You find as early as possible the indicators that you're doing the right things. Start with a core belief, and unless something changes your mind, you continue to believe in it and look for early indicators of future success.
主持人Interviewer:What are some of those early indicators that have been used by product teams at Nvidia?
Jensen Huang:There are all kinds! I saw a paper on deep learning before I even knew what deep learning was. Then, I met some researchers who needed our help creating a domain-specific language for their deep learning algorithms to work on our processors. We created cuDNN, essentially the SQL for neural network computing. We created this language specifically for deep learning, kind of like the OpenGL of deep learning. They needed it to express their math, and they didn't understand CUDA but did understand deep learning, so we created this middle ground for them.
Interviewer:Even though there were zero financial returns at the beginning, you were willing to do this?
Jensen Huang:Yes. This is one of the great strengths of our company: we're willing to do something even though the financial returns are completely non-existent or very far out. We ask ourselves: is this worthy work to do? Does this advance a field of science somewhere that matters? We find inspiration not from the size of a market but from the importance of the work. The importance of the work is the early indicators of a future market. Nobody needs to write a business case on it. The only question is: is this important work, and if we didn't do it, would it happen without us?
Interviewer:These selections that you've made have paid huge dividends, but you've had to steer the company through some very challenging times, like when it lost 80% of its market cap during the financial crisis because Wall Street didn't believe in your bet on machine learning. In times like these, how do you steer the company and KEEP the employees motivated?
Jensen Huang:My reaction during that time is the same reaction I had about this week. My pulse was exactly the same. You just go back to what you believe. You gut check yourself. What are the most important things? You check them off. Family loves me, okay? Check. You go back to your core, then go back to work, and keep the company focused on the core. Did something change? The stock price changed, but did anything else change? The physics? Gravity? No? Then you keep going.
Interviewer:Speaking with your employees, they say that you try to avoid public speaking. They said that your leadership style is very engaged. Can you tell us why you've purposefully designed such a flat organization?
Jensen Huang:No task is beneath me. I used to be a dishwasher! You can't show me a task that's beneath me. If you send me something and you want my input on it, I can be of service to you and share how I reason through it. By reasoning, I empower people. I show people how to reason through strategy, forecasting, and problem-solving. It takes a lot of energy, but I feel rewarded by the process. A CEO should have the most direct reports by definition because the people who report directly to the CEO require the least amount of management.
Jensen Huang:My job is to create the conditions by which you can do your life's work. That condition is empowerment, which comes from understanding. You need to be informed, and the best way to achieve that is to have as few layers of information between us as possible. That's why I reason through things openly. This creates a highly empowered organization. Nvidia's 30,000 people are the smallest large company in the world, but every employee is empowered and making smart decisions every day. They understand the context because I'm transparent with them.
