Summarize this content to 2000 words in 6 paragraphs in Arabic
You can enable subtitles (captions) in the video player
These are amazing times because we’re at the beginning of a new industrial revolution. This is the heart of US exceptionalism. Nvidia has the Midas touch in Silicon Valley. Revenue growth year on year of 200 per cent or more. The emergence of DeepSeek now putting pressure on the US tech names. Some people have said that DeepSeek has shown that the US is behind on AI, and that China is creeping forward. It turns out that you can possibly build a model which is almost as good as the very best in the business without having access to Nvidia’s best chips, and made at a fraction of the cost. Artificial intelligence has become so advanced it has now surpassed human performance on several basic tasks. Over the past few years we’ve seen a real revolution in the artificial intelligence industry. A lot of companies and labs got set up to develop this technology, and they found that it was really important to scale what they were doing, to apply more computing power, to ingest more data, to come up with smarter algorithms to do that. And what they found was that Nvidia’s chips were fantastically flexible, and good at performing all of the functions that they really needed to build AI. It’s basically had a second life, which has surged it to one of the most important and valuable companies in the world. And they’ve become the hottest commodities in tech. Let’s talk about DeepSeek because it is mind-blowing, and it is shaking this entire industry to its core. It turns out that you can possibly build a model, which is almost as good as the very best in the business, without having access to Nvidia’s best chips, which don’t make their way to China because of export controls. So DeepSeek seems to have done what it can with what it has, with quite remarkable results. The release of DeepSeek AI from a Chinese company should be a wake-up call for our industries that we need to be laser-focused on competing to win. DeepSeek is a Chinese AI company which burst onto the scene. It was relatively unknown before, and it has these capable models, which are beating a lot of the rival companies or matching a lot of their capabilities. What seems to have happened is that DeepSeek has found workarounds. They’ve taken shortcuts. It’s found new ways to train the algorithms to get them to a similar level as OpenAI. Now, we don’t fully know exactly where DeepSeek’s technology came from. That’s going to emerge over the coming weeks and months, if at all. The sort of frightening thing for the US competitors is that if the export controls were effective, it’s almost like what’s happened is that DeepSeek has found a way to make apple pie without using apples. After DeepSeek came out with their reasoning model the US stock market clearly freaked out, and $1tn was wiped off the value of the leading US tech firms in one day. The phrase of our time seems to be ‘the Sputnik moment’. Originally, when the Soviets put up the first satellite in 1957 the Americans panicked and thought that they were getting a technological edge over them, and therefore invested massively. Some people have been talking about DeepSeek and the development of its reasoning model as another Sputnik moment for America. AI is a tool, it’s entertaining, it does things for businesses, but it’s also a weapon. And it’s really important to remember that when you’re thinking of the response to all of this from DC. Lots of AI companies, including OpenAI and Meta Platforms are questing towards this thing called artificial general intelligence, which is the point at which AI becomes as intelligent as a real human being. This is kind of the holy grail, and whoever gets there first, whichever country gets there first, is going to have a formidable advantage over its trade partners and rivals. The British mathematician IJ Good, who worked at Bletchley Park and was one of the pioneers of computing technology, came up with this idea of humankind’s last invention, by which he meant an ultra-intelligent machine that would be able to invent everything that we could invent. So we reach a certain point of development where we effectively hand over the baton to an electronic intelligence, that then can invent everything that we would have been capable of inventing ourselves only in faster and more capable ways. Now, this has been a somewhat theoretical discussion and remains a theoretical discussion, but that really is where people are hoping to get with AGI. In this race between the US and China to attain hegemony in AI, the US has been looking to kind of hamstring the Chinese. They don’t think it’s very sensible for them to export their state-of-the-art chips to enable the Chinese to develop very powerful AI systems. Our export controls, not backed by tariffs, are like a whack-a-mole model, where they get prevented over here and China figures out a way around it over there. We’ve got to find a way to back our export controls with tariff model, so that we tell China, you are… you think we are your most important trading partner. When we say no, the answer is no. It’s a respect thing. First, Donald Trump and then Joe Biden have imposed quite strict export restrictions, both on the chips themselves and the equipment that is used to manufacture those chips. And this has undoubtedly had a big impact on the development of the Chinese chip sector and the AI industry as well. But perversely, that restraint has in fact acted as a great spur to innovation within China. Some people have said that DeepSeek has shown that the US is behind on AI, and that China is creeping forward. I don’t think China has ever been behind on AI. It’s something that, that country has been working on for a very long time. It has a huge amount of AI resources and talent, and it’s got to be seen as a credible competitor to the US. There was a whole narrative that the only way you could play in AI was to invest massive amounts of money in computing and data and very smart researchers. And we saw recently the Stargate announcement where Donald Trump was standing next to Larry Ellison from Oracle, and Sam Altman from OpenAI, and Masayoshi Son from SoftBank, who had committed to invest as much as $500bn, although the initial tranche was only for $100bn, in developing these massive data centres that were going to be the infrastructure for this latest AI revolution. But that really only added to the extraordinary impact that DeepSeek had later that week, when it became apparent that other people were experimenting with different AI methodologies on a far cheaper way, and with far less compute power, and coming up with extremely powerful models. We also don’t yet know how DeepSeek will progress. Will its models get much better? Is this as good as it gets? Will we find out new information about the way it was produced that changes the story somewhat? We don’t yet know. Nvidia’s Blackwell chips, which it says are in insane demand, may not be the only way to get to a really high standard of AI. So it wouldn’t be surprising if Nvidia’s valuation takes a knock and doesn’t entirely come back. But do we need Nvidia? Absolutely, yes. Are its chips the best in the business? Undeniably so. They also make the software which allows you to control and organise the server farms. So Nvidia really does have a bit of a chokehold or a stranglehold on the industry at the moment. There’s a lot of money to be made in the video game business. Last year $5.5bn worth. The two biggest video game makers are soon going to bring a new game to town, one that runs on compact discs. The earliest insight that Nvidia’s founders had was that computer graphics were going to be a big thing. That was something that deserved its own processor. And this was a time when Intel ruled the world, partnership with Microsoft on Windows. Wintel was the dominant computing platform and would continue to be for decades. Video games at that time, in the early 90s, were starting to become a big thing. Nintendo and Sega had all started but these were not very graphically rich things. So the insight was that this is something that you can build a dedicated chip for, because they were catering to an audience of video gamers who just wanted ever faster chips for ever better graphics and increasingly frenetic multiplayer battle. Those games gave Nvidia a reason to exist, and to keep existing, and set up the customer base for a type of chip that otherwise might have just been absorbed into the standard Intel CPU. The early history of Nvidia is not one of immediate success. It’s one of struggle and sacrifice. And Jensen really, I think, has retained that philosophy, that there is this kind of Hobbesian world you’re operating in. To this day, I use the word… the phrase ‘pain and suffering’ inside our company with great glee. And the reason… and I mean that. You know, boy, this is going to cause a lot of pain and suffering. And I mean that in a happy way. Because you want to train, you want to refine the character of your company. You want… you want greatness out of them. And greatness is not intelligence, as you know. Greatness comes from character, and character isn’t formed out of smart people. It’s formed out of people who suffered. Jensen’s a very interesting character. He moved to the US from Taiwan when he was nine. He went to Oregon State University, and then he went on to Stanford, where he studied engineering. Because of his engineering background he clearly has a deep understanding of the product itself. Our brand new GeForce RTX 50 Series Blackwell architecture, the GPU is just a beast. I think a big part of what makes Jensen interesting in Silicon Valley today is he is the last of the pre-dotcom generation of founder CEOs. He is out there living the rock star life, appearing on stage with everybody that he can think of, even some of his biggest competitors and companies that want to be his biggest competitors, like Masa Son from SoftBank or Satya Nadella at Microsoft. He’s up there on stage, shaking hands and joking. It was just bizarre when it was happening, just how quickly this company was transformed into being one of the superstars. And Jensen Huang, the CEO as well, became one of the most globally recognisable people in business and finance. This is not a company with a very traditional hierarchy. Nvidia frankly sounds quite chaotic to work for, because Jensen can walk past any engineer’s desk and will know exactly what they’re working on, will know enough about what they’re working on to make some pretty detailed questions about it. He has this ability to reach down into the organisation and does not want to do meetings, does not want to do appraisals, does not want to share information with his top lieutenants. He wants everyone to know what everyone is working on. So his management style is very, very different to most American businesses. Maybe there’s a slight sort of chaos element in a lot of startups, but normally by the time you get to a $3tn valuation, things have calmed down a little bit. Not so much at Nvidia. I never finished my business plan. I know it. I know it. We never finished a business plan, never could figure out how to finish a business plan, to tell you the truth. Nvidia kind of accidentally stumbled into AI. Its chips were made for rendering video games graphics, and it was by accident that they were discovered to be hugely powerful and useful for designing AI systems. Over a decade ago, researchers basically figured out that if you use these GPUs and try to deploy AI models onto them that they could scale. And that was really an advancement for the theory of neural networks, which is what underlies a lot of the AI that we have today. It’s a system where AI kind of operates like the human brain. The real question around Nvidia is what’s happened in two years that’s taken it from a roughly $300bn company to a $3tn company, that’s potentially one of the most valuable in the world, it’s vying with Apple and Microsoft for the top spot. Hundreds of percent multiples increase in just a short space just by the surge in its market capitalisation and share price, the demand for its products. There really isn’t anything better than what Nvidia are offering in terms of GPUs for this at the moment. What we are now seeing, though, is an attempt from everybody, not just the big rivals like Google and Meta and Amazon, to reduce this reliance on Nvidia, to make their own chips, to design their own servers, bring in their own software. One, because they want a bigger share of the pie for themselves, but also they want, as a secondary measure, they want control of their own destinies. They do not want to have to go cap in hand to Nvidia when they need more chips. They don’t want somebody else in control of which of their rivals gets the most chips. You’re also seeing companies like AMD, who are trying to make their own rival chips, which can be cheaper and more cost effective. And you have this supply and demand problem where Nvidia is struggling to meet all of the demand for its chips. So if another company can come in, offer similar capabilities at a better price, then companies might decide to go with them. I’m an economist. I have great confidence in the market system. It gets some things wrong, it gets some things right, but the competitive process is a very healthy one for economic growth. But that competition becomes quite strained when one or two companies dominate everything. And we’ve become accustomed to that in the information age. And Nvidia is perhaps pushing that to the next level. We have plenty of examples from history where innovations in some domain create scarcities in others. The British Industrial Revolution’s early phase was characterised by tremendous advances in spinning, which suddenly made weaving very scarce. So it led to a huge increase in weaver wages. But then what happened is that the power looms came, and weaver wages fell by almost half. New, creative destruction can actually change the dynamics at one fell swoop. And one might think this is not out of the question for Nvidia. Somebody else may come up with a chip that is far superior, and then that could change dynamics completely. In the aftermath of the markets being impacted by DeepSeek, Satya Nadella, the chief executive of Microsoft, referenced something called Jevons paradox, and that was essentially the theory that coal consumption in the Industrial Revolution, despite increases in energy efficiencies, didn’t decrease. The demand for it only increased, and that’s because people were consuming more and more energy than ever. And so the counter point for Nvidia that it’s hopeful for is that people will continue to need chips for AI, whether that’s for training or actually whether it’s for something called inference, which is running the models. As consumers start to use AI more and more, the argument is we will need the infrastructure there to be able to meet demand for these products. The atmosphere out here on the West Coast is feverish. People that have been out here for years say it reminds them of the dotcom boom or the real take-off of consumer mobile phones. There is huge optimism about both the potential of AI to transform business and, as always, the potential for AI to make a lot of people a lot of money. It’s really been kind of the shot in the arm that San Francisco as a city and the wider Bay Area, including Palo Alto, kind of needed after a few dire years, some difficult years during the pandemic. Suddenly, they had this new technology that was going to touch every part of business and human society to kind of organise around and to compete with. It’s really injected a sense of optimism, urgency, but increased competition. Machine learning and artificial intelligence has been around a long time, decades. And it’s gone through a number of, what they called AI winters, where people would get very excited, and they’d do a lot of stuff, and then they would realise that it just wasn’t viable. And Nvidia, they certainly recognised the opportunity when it presented itself. And they went whole hog into this in developing not just the hardware, but the software and the ecosystem they’d already started to build around graphics processing for games, consoles, to build it around artificial intelligence, when it became very clear that the math that was used for both and the types of calculations that were used for both were very, very similar. So they reached out with both hands and kind of grabbed this opportunity. And they’ve been building parts for this for many, many years. Now, it’s only recently that we finally hit the inflexion point. Nvidia’s chips are being used by OpenAI, who have a long history of using these GPUs and developing AI systems. And they were building these large language models which had really AI at scale that could speak in a conversational way. And that’s what then became the chatbot that we know as ChatGPT, which in 2022 just exploded into public consciousness where people for the first time could really interact with AI systems, and see how advanced they had got, where they could talk to the system, and it would respond in a way that was like talking to a human. When you look at the insane growth of Nvidia, that has a sort of rising tide effect. So there is a whole ecosystem of companies whose fortunes have somewhat been tied to and also been sort of feeding Nvidia’s success. There are certain companies which Nvidia has developed partnerships with over the years, some of these more hybrid cloud companies who use Nvidia’s chips, and then sort of rent out the capacity system of using an Nvidia chip. Suddenly, they realised they were sat on this lucrative stack of hardware that could be sold or rented out for much more money to people suffering from a lack of capacity in the AI industry. So they were like, we can benefit from this. So they’ve pivoted their companies, and people like Lambda and Cohere have been raising billions in financing, they’re raising debt from some of the largest Wall Street banks, and they’re even raising money from Nvidia, It’s kind of a circular financing arrangement. Nvidia is providing them financing to buy some of their… more of their chips. We have a really close relationship with Nvidia. We’re a critical vendor for them because we not only just sell GPUs, but we offer services that make their GPUs more accessible, whether it’s just really good support or access to GPUs via our cloud. Nvidia views this us like a key partner because we make their product more accessible. In turn, they are willing to give us allocation. So we’re building a cloud for AI. Our job is to make it easy for people who are doing AI research to do their research. A lot of that focus is on infrastructure. So AI research requires a lot of computation to do well. And what used to be kind of like the purview of national labs, where you’d build these giant supercomputers, is now something that’s been brought into AI, and running that infrastructure is enormously challenging. Lambda is a case of a company that’s managed to raise millions of dollars in loans using the chips as collateral. And also has seen very, very speedy funding rounds and succeeding each other very fast and sort of off-scale growth over the last year or two. It’s very hard to predict whether something is a bubble, but certainly there is a huge amount of interest in AI at the moment. And it’s not clear when that interest is going to stop. It’s also not clear when these companies are going to start making money. There was a particular note from a Goldman Sachs’ head of research, which attracted a lot of attention, basically identifying a single-digit billions revenue opportunity versus hundreds of billions being invested per year. Microsoft, Meta, Google, they are spending collectively every six months, more than $100bn, largely on buying data centres, filling them with chips and networking equipment and servers, to kind of feed this insatiable demand for AI-related computing power. This has been compared to the next industrial revolution. And certainly you’re hearing lots of tech chief executives talk about AI in comparison to lots of theories of the industrial revolution, as well. And it’s certainly a disruptive technology. There’s lots of talk about the types of sectors this is going to impact, and the displacement of jobs, which might be automated out of the workplace by this technology. A lot of people out in Silicon Valley telling me that this is very similar to the dotcom bubble and boom, and that this technology is going to be just as influential as the internet, if not more. And I think we’re kind of at the stage now where we have the concept, and we have the technology, and the capabilities, but we don’t have that sticky app yet. We don’t have that use case or use cases that are really changing the world. The bear case for this is that the corporate uses and the consumer uses that people willing to pay for chatbots are simply not as great or persuasive or as well-developed as people think and hope they are at the moment. We’re not going to see every call centre replaced with just a super smart AI chatbots that can do the job more effectively, much faster and crucially for companies, much cheaper. What we need to see, I think, to justify some of the valuations of the companies and the money going in is some real-world use cases, real-world revenue being generated that will kind of calm people down a bit. And if that doesn’t start to come through I would say then it’s going to start to be a bit concerning. And a lot of companies are going to have to make really tough decisions on their spending and job cuts. Please raise your right hand and repeat after me. I, Donald John Trump, do solemnly swear. I, Donald John Trump, do solemnly swear. That I will faithfully execute… Trump’s a bit of a wild card in two directions for Nvidia. Trump has talked about Taiwan as stealing the US semiconductor industry. They took our business away. We should have stopped them. We should have taxed them. We should have tarifffed them. Nvidia doesn’t operate in a vacuum. I think that’s really important to be aware of. They don’t actually make the chips themselves. They design the chips. They send the designs to, what’s called, a fab. It’s just like a manufacturer. North of like 90 per cent of the most powerful chips in the world are produced by one company, TSMC. It’s a huge geopolitical question for the future about the need to retain access to that chip supply, because it’s possibly the biggest vulnerability for Nvidia is you’re making billions and billions of dollars off chips, which you design in the US. But then to actually bring them to customers, you need to ship them from Taiwan. If TSMC’s operations are disrupted by an earthquake, which, you know, happened quite recently, and things were fine, but might not be next time, or by some kind of incursion by China, whether that’s a blockade, or something, you know, slightly hotter than that, this is something that would grind the AI revolution to a halt very, very quickly. These are becoming geopolitical issues. This is an incredibly geopolitically sensitive area of the world. It’s kind of like China’s Ukraine, if you’re looking at it from a kind of the Russia angle. A lot of people are very concerned about what happens. The US is in the process of engaging in this kind of historic investment by the US federal government in building US chip manufacturing capacity. So the success of that project will be really critical for Nvidia in the long term. And just yesterday Taiwan Semiconductor, the biggest in the world, most powerful in the world, has a tremendous amount – 97 per cent of the market – announced a $165bn investment to build the most powerful chips on Earth right here in the USA. And we’re not giving them any money. Your CHIPS Act is a horrible, horrible thing. We give hundreds of billions of dollars, and it doesn’t mean a thing. They take our money, and they don’t spend it. The Industrial Revolution was the beginning of a very painful process. It wasn’t all smooth sailing. It’s not like, oh, yeah, the Industrial Revolution came, machines started helping workers and consumers, and everybody suddenly became much wealthier. It took many new technologies, many new companies, a redirection of the overall scientific and industrial effort of the entrepreneurial and the scientific class in Britain and elsewhere. So if Jensen Huang is anticipating all of these institutional and technological changes, I think he may be right. But it also means that we’re going to have a very long and very hard several decades ahead of us. There’s a question for the government about how it wants to play this, and how AI companies in the US are really going to compete with the power that Chinese AI has. We’ve already seen these export controls on Nvidia’s shipments into China, and they have to ship versions of their chips for China which are pretty hobbled compared to the types of equipment that you can deploy in the US. And yet, DeepSeek claimed that they managed to develop this technology on H800s, which are more inferior chips and permitted under these export controls. So they didn’t do anything nefarious, but were still able to generate systems with similar capabilities to their competitors in the US. And that’s created concerns that perhaps the export controls aren’t working, maybe they need to be weakened because they’re redundant, or maybe they need to be tightened, and we need to prevent more spread of these highly powerful chips going to China. There’s no doubt that both US and Chinese companies are pawns in a big geopolitical game at the moment. Your classic economist would say that there would be far faster development of the industry as a whole if there was free trade between the US and China, and both were able to concentrate on their fields of expertise, and there was a free exchange of views and methodologies between the two sides. But we’re not seeing that. And in fact, it’s quite the reverse. We must be able to build the chips and semiconductors that we need right here in American factories with American skill and American labour. And that’s exactly what we’re doing. We’re seeing a separation between a blue supply chain and a red supply chain. And so this very integrated industry that we had seen develop in the postwar era has now been pulling apart. And you are seeing US and China developing their industries very separately.
