Guideposts: DeepSeek: Good for Nvidia, Great for America | | by George Gilder and Richard Vigilante
01/29/2025 | | | SPONSORED CONTENT Elon Musk Dreams, Mode Mobile Delivers As Elon Musk puts it, "Apple used to really bring out products that would blow people's minds." Those days are long gone, and a new smartphone company is stepping up to deliver the mind-blowing moments we've been missing. They've just been granted the stock ticker $MODE by the Nasdaq, and you can still make an investment in their pre-IPO offering.
Turning smartphones from an expense into an income stream, Mode has already helped users earn and save an eye-popping $325M+ and seen an astonishing 32,481% revenue growth rate over three years. Uber did it to taxis, Airbnb did it to hotels…And now, Mode Mobile is doing it to the $1 trillion smartphone industry.
Don't miss your chance to invest in the next big disruption to smartphones.
Join 30,000+ shareholders and invest at $0.26/share today.
This is a paid advertisement for Mode Mobile Regulation A offering. Please read the offering circular and related risks at https://invest.modemobile.com/ | | | Reports that Chinese start-up DeepSeek can train ChatGPT-like AI models for 5% of the costs incurred by OpenAI —$5 million versus $100 million— cratered U.S. tech stocks, and most of all Nvidia Corp. (NASDAQ: NVDA), which lost $600 billion in market cap in one trading session, a global record.
Even more dramatic, DeepSeek's cost to users may be as little as 1% of the cost of using ChatGPT.
Our Gilder Technology Report will do a deep dive on the implications for investors, including a special "Hotline" tomorrow updating our valuations and recommendations on NVDIA. To get that report tomorrow, subscribe here now.
This coming Friday, our sister publication, George Gilder's Moonshots, takes an even deeper dive into the technology itself and evaluates the hype against the reality. To get that issue (and other reports on the future of AI) in 48 hours, go here NOW.
One thing is clear: DeepSeek is not a mere incremental improvement—it's a fundamental reimagining of AI development. It transforms advanced artificial intelligence from a luxury of tech giants into a tool accessible to smaller players and startups. As with the democratizing effects of personal computers and cloud computing, DeepSeek's architecture could reshape the entire landscape of AI innovation.
DeepSeek's innovation is not new hardware. It is hugely more efficient use of existing hardware, which was already the most powerful trend in AI development. As we discuss in greater detail in the upcoming George Gilder's Moonshots, the innovation is primarily in the algorithms, which economize on what they demand of the hardware.
One of several ways this is done is to employ "parameters" (variables the model adjusts to make decisions) more selectively. The DeepSeek model contains more than 650 billion parameters but activates a mere 5% of them at any time, based on the task. Biology parameters are activated for scientific questions, while coding parameters are employed for programming challenges. | | The Biggest Winner of the AI Boom Isn't Nvidia… Nvidia (NVDA) has soared more than 1,700% over the past 5 years. For investors who missed out on the profits, America's #1 Futurist says AI is converging with a 'miracle material' right now, and one company, leading the way, could see its shares post 10X gains… | | | Along with other techniques, this drastically reduces computational demands and energy consumption. Historically, developing state-of-the-art models like GPT-4 required tens of thousands of GPUs. DeepSeek has reportedly reduced hardware requirements to just 2,000 GPUs, making it possible to train advanced models on affordable setups, such as PlayStation 5 clusters, liberating AI from dependence on NVIDIA's most elite and expensive hardware.
That declaration of independence is what crashed both Nvidia's stock and tech indexes. The history of the semiconductor industry says Mr. Market got it wrong again.
Semiconductors are the leading example in all history of Jevons' Paradox. Jevons discovered that when a resource is rendered more efficient, we use more of it, often so much more that total spending on the resource rises. When the world was lighted by fire, it was a very dark place. As Nobel laureate William Nordhaus famously pointed out as we progressed from candles to oil lamps to incandescent lights and now LEDs, the cost of lighting dropped by 99.97%, yet we buy more of it than ever.
That's the story of the microchip, except that chips have moved at a faster pace. The cost of a transistor has dropped by several million percent in 70 years while annual spending on semiconductors has grown from less than a few hundred million to nearly $700 billion. The cheaper computation became, the more it was demanded.
Economic progress depends on wasting what is abundant and conserving what is scarce. Today, the AI industry is desperately striving to conserve spending on computation. As the combination of more efficient algorithms and more efficient hardware makes computation more abundant, expect more AI computing to be demanded, not less.
We know that the hardware has gotten dramatically cheaper: the power of GPUs per dollar has been progressing faster than Moore's Law pace. The spread of practical AI is based on this reduction in price. So, paradoxically, is Nvidia's startling growth in revenue. The H200 is minting money for Jensen Huang's company because, per computation, it is cheaper than the H100.
By drastically lowering the costs and hardware demands of AI, DeepSeek is more likely to spur increased demand for high-performance GPUs, such as those produced by NVIDIA. As for "commercial" lower performance GPUs, they were how NVDA made its living as little as three years ago. As AI adoption accelerates across industries due to reduced barriers, the need for computational power to support this growth could reinforce NVIDIA's dominant position. | | Search Less, Profit More: January A.I. Options Scanning Did you know there are over 10 million tradable assets worldwide?
Finding the right opportunities can feel like searching for one grain of gold in a vast desert.
Imagine having a tool that acts like a metal detector, cutting through the noise and pinpointing high-potential trades with precision.
With A.I., you spend less time searching and more time focusing on what really matters — making profitable decisions.
Join the January A.I. Options Scanning Session. | | | The most important question for Americans, however, is how could a small Chinese start-up achieve this in the face of U.S. sanctions depriving Chinese companies of leading-edge AI hardware, sanctions specifically intended to retard China's AI efforts?
The best answer: It was the U.S. sanctions that made it possible. DeepSeek should thank Joe Biden.
Necessity is the mother of invention. Israel has the world's best anti-missile defense because the Israelis had to figure out how to do it on the cheap. Israel has the world's best main battle tank because it could not afford $12 million per tank for the most advanced armor. Instead, the Israelis simply put the engine in the front; the engine supplements the armor. The result is that crews of the $3.5 million Merkava tank are better protected than the crews of the U.S. Abrams.
In the United States, aspiring AI companies are all trying to lower costs by economizing on hardware. That's a good incentive to innovate. As we detail in the Gilder Technology Report and George Gilder's Moonshots, much innovation is happening. The Chinese had an even better incentive: they could not buy the elite hardware at any price.
China is home to the greatest number and some of the greatest electronics engineers and software creators in the world. It's time to stop trying to knee-cap them. Let's admit they have much to teach us. Let's collaborate and compete rather than trying to punish China by holding our breath till we turn blue.
Sincerely,
George Gilder, Richard Vigilante, Steve Waite, John Schroeter, and Robert Castellano
Editors, Gilder's Guideposts, Technology Report, Technology Report Pro, Moonshots, and Private Reserve | | About George Gilder:
 George Gilder is the most knowledgeable man in America when it comes to the future of technology and its impact on our lives. He’s an established investor, bestselling author, and economist with an uncanny ability to foresee how new breakthroughs will play out, years in advance. George and his team are the editors of Gilder Technology Report, Gilder Technology Report Pro, Moonshots and Private Reserve. | | | | | |
Visit our YouTube Channel - Eagle Investing Network
No comments:
Post a Comment