Malcolm Penn, CEO of Future Horizons and Europe's leading semiconductor analyst, has written an insight into the NVIDIA phenomenon.
At what can only be described as "the right place, the right time," Penn writes, NVIDIA's meteoric rise over the past year has rewritten the history of the chip market.
The fuse was lit on November 30, 2022, when OpenAI released an early demo of ChatGPT in the unsuspecting world.
Overnight, the chatbot redefined the goals of artificial intelligence (AI), proving that machines can indeed "learn" the intricacies of human language, interactions, context, nuances, and even humor. It quickly went viral on social **, with users sharing examples of its features, from travel planning to writing pop songs and computers**.
Within five days of its release, the app had over 1 million registered users and is now widely used in a variety of different industries, including:
Customer Service: Automate responses to common inquiries. Education: Create a personalized intelligent tutoring system. Content creation: Everything from generating ideas to writing articles. Enterprise: Used to generate email drafts or write**. Healthcare: From clinical decision support, record keeping, and analysis of medical literature Entertainment: Generate game storylines, movie scripts, and dialogue. It has also had a profound impact on the evolution of artificial intelligence. In the midst of a flood of hope, hype, and hysteria not seen since the internet boom in 2000, AI is at the forefront and heart of every company's dream.
However, no company has benefited more from the AI boom than NVIDIA. Since the beginning of 2023, the company's share price has **reached nearly 400%, and although its 2023 net profit is higher than NVIDIA's 2023 revenue, its market capitalization has reached 1At $8 trillion, it has surpassed Alphabet to become the third most valuable company in the United States, after Microsoft and Apple.
At the moment, only a few brave people dare to bet on Nvidia's market capitalization further, as investors are fascinated by their dreams of a future for AI supremacy.
Every company has its own day
NVIDIA was founded in April 1993 by three American computer scientists: Jensen Huang, formerly LSI Logic and AMD; Chris Malachowsky, former Sun Microsystems employee; and Curtis Priem, a former IBM and Sun Microsystems employee.
Given its combined background in graphics and processors, the startup's focus is on graphics-based computing and the gaming market.
Over the next decade, NVIDIA cemented its position as the powerhouse of gaming console chips with its leading graphics processing unit (GPU) technology.
The world was dominated by computer processing unit (CPU) devices, Intel was the market leader at the time, and Arm was just entering the fray, and these devices were at the heart of compute-driven tasks.
CPU-driven tasks require faster and more complex processors, while GPUs are tasked with making movements and animations more realistic and natural.
CPU devices are typically serial functional structures, while GPUs take advantage of parallelism to perform many much simpler tasks, breaking down a single problem (such as waving a hand) into sub-parts, and then solving them simultaneously using multiple simple processors.
The company's focus on parallel computing technology has not only helped NVIDIA move from DotThe recovery from the collapse of the com market also helped the company establish a leading position in the field of artificial intelligence, which was still far away at the time. Whether it's luck, good fortune, or forward-looking, the company's parallel processing technology is exactly what the computing speed and power performance required to develop and run complex AI applications.
In 2006, NVIDIA introduced the Unified Computing Device Architecture (CUDA) platform, which ultimately proved a major innovation in GPU-based parallel computing technology. But it wasn't until 2022, when word broke that OpenAI's ChatGPT platform was using 10,000 NVIDIA GPUs, that NVIDIA's investment in AI paid off.
As a result, rival platforms snapped up NVIDIA chips, and demand quickly far outpaced, and the number of available chips skyrocketed, and Nvidia's sales soared. As investors scramble to buy nvidia, nvidia's share price has soared similarly.
What's next?
With a nearly 50% gain so far this year, dwarfing the S&P 500's 5% gain, the industry has run out of superlatives to describe the giant of AI chips.
One of NVIDIA's attractions is also the enthusiasm around AI, as companies are scrambling to find opportunities to build their AI capabilities in order to improve their market position or run their businesses more efficiently.
However, very few companies, especially from a corporate perspective, are able to realistically say what they want to be widely used in, which is a prerequisite for AI to prove that it's more than just hype.
NVIDIA's future lies in its ability to continue to bring innovative hardware and software solutions to market to solidify its current dominance in the GPU space and its potentially transformative impact on the future development of AI.
Clearly, the company is no slouch when it comes to innovation, and it's just as clear that it has an impressive head start over its traditional semiconductor competitors.
It takes at least two years to design such a complex chip and bring it to market, so even if a potential competitor starts moving right away, it will still have to wait a year until the first new product appears on the market, and four to six months before the first products are shipped.
Given its first-mover advantage, NVIDIA will release two enhanced versions of its current parts roadmap by then. Catching up with market leaders is never a good place. At most, it is two steps forward and one and a half steps back.
Despite this, AMD, NVIDIA's strongest semiconductor competitor, is already well on its way to "following NVIDIA" with the Mi300 chip. The 153 billion transistor chip, launched in December last year, has 192GB of memory and a memory bandwidth of 53TB, which is about the top of NVIDIA's H100 AI chip. 4 and 16 times.
As impressive as this may be, NVIDIA's lead in software can be more difficult to address. It has been investing in CUDA since the mid-2000s, and AI developers love the way the platform allows them to fine-tune GPU performance.
AMD is counting on breaking NVIDIA's shackles in this regard by open-sourcing its competitive ROCM software platform and providing the tools to convert CUDA programs to ROCM.
Watch out for the back?
However, the bigger competitive threat may not come from companies like AMD, but from NVIDIA's high-tech customers. Tech giants from Alphabet, Amazon, Meta and Microsoft are busy designing their own AI accelerator chips, desperate to break away from NVIDIA's market dominance, which allows it to charge sky-high prices and quantitative chips**.
The chips are also tailored to the task at hand, delivering higher performance and cost savings at the end system level, an approach that has allowed Apple to achieve such good results with its self-designed iPhone chips.
Once this transformation is complete, there is no way to recover, and these OEM customers will never come back.
The only way out is to fall?
Faced with the risk of rain during the parade, we believe there are two areas of risk.
First, the current AI hype and potential overinvestment has to do with the 2000 dotcom booms and busts are the same, and now, as then, no one will believe that a depression is coming, and no one will foresee what will trigger it.
In any case, technological transformation is always a few years longer than the thriving record share price indicates. And, when a tech CEO asks for trillions rather than billions of dollars when it comes to raising money for an AI chip project, you know deep down that the industry may be a little too hot.
Second, the current optimism in market valuations goes against the fundamentals of the market. According to rough calculations by technology-focused analysts at Chameleon Capital Holdings Limited, NVIDIA's valuation means the company will maintain a monopoly-like operating margin of 55% over the next decade, while sales will increase tenfold, from the current level of $60 billion to more than $600 billion.
With this in mind, the entire chip industry generated $527 billion in sales in 2023, and this number is expected to grow to 1 by 2034$4 trillion.
Admittedly, NVIDIA has indeed achieved this level of growth over the past decade, but that has come at a much lower level of $4 billion in sales, and its current level of profitability is a recent phenomenon associated with pushing prices in response to overwhelming demand and product shortages.
The previous profit level was between 12% and 37%, well below the required stable level of 55%.
For the good times to continue, the law of large numbers must cease to apply, and competition, innovation, and pricing pressures won't emerge until at least the mid-2030s.
A living legend
NVIDIA deservedly got the "right time and place" moment ......It's been around for 30 years, and the history of the chip industry flourished in legendary moments like these.
In the long run, AI will undoubtedly change many industries and the way people work, but the path to this ultimate goal is currently shrouded in ecstatic hype and hysteria.
AI isn't a product per se, just like an iPhone or laptop, you can't log in to Amazon or Etsy and buy AI. Arguably, products like ChatGPT are more of a human plagiarism than intelligence.
That's what the chip industry does best, and it's the ...... that has driven growth over the past 70 yearsBut yet hasn't a chip market taken off based on a $40,000 IC!