What is the most fierce price increase in 2023?
There is no doubt that it is NVIDIA's computing power chips, how popular they are, and the graphics cards downstream of the GPU are bought out, whether it is new or old, the inventory is directly cleared.
According to statistics, NVIDIA's RTX 3060 has been fried from the official 2499 yuan to nearly 6000 yuan, an increase of more than 2 times, and it is often out of stock, and most of them need to make an appointment to buy it.
The RTX 3070, whose official original price is 3899, has soared all the way to more than 8000**. The most fierce RTX 4090 graphics card has risen to 50,000 yuan.
Until now, I can't even grab the graphics card, and the technology giants of the major manufacturers have begun to target the downstream of the "graphics card", the server!
This is not on March 1, 2024, Dell Technologies rose by more than 31% to hit a record high since the IPO in the United States in 2019, because the technology giant is ready to remove it from the server for the GPU!
Then in the case of such a shortage of GPUs, chips with better average performance than GPUs, FPGAs known as universal chips, are put on the agenda.
What is an FPGA?
FPGAs, called Field Programmable Gate Arrays, are semi-customized, programmable integrated circuits.
Common chips are fixed after they are manufactured, and users cannot make any modifications to their hardware functions.
After the FPGA chip is manufactured, users can assemble the chip through chip design software (EDA) according to their own needs and ideas, similar to the building blocks in integrated circuits, and build them into different shapes to meet the application needs of different scenarios.
Therefore, the FPGA chip is also known as the "all-purpose" chip, which has become a "customized" chip for each user.
In addition, the FPGA chip also responds quickly, and the computing efficiency is higher than that of the GPU. Since the FPGA chip adopts the "hardware underlying architecture", the information is transmitted in memory "without instructions".
Both CPUs and GPUs use the "von Neumann structure", and information transmission needs to be executed by "instructions". Therefore, FPGA chips are more efficient than CPUs and GPUs, and FPGA chips have a faster response speed when processing massive amounts of data.
Therefore, FPGA chips are more suitable for deep learning of large AI models, and FPGA chips also have the advantages of high flexibility, low application development cost, and short time to market.
What is the upstream and downstream of FPGA chips?
The upstream of FPGA chips, like all chips, is formed by wafer processing, packaging and testing. The differences between FPGA chips are mainly focused on the downstream direction.
FPGA chips benefit from a unique architecture and have incomparable advantages over other coprocessors such as CPUs and GPUs in terms of flexibility. Therefore, the downstream application scenarios of FPGAs are very rich.
At present, it has been widely used in various fields such as industrial control, network communication, consumer electronics, data center, automotive electronics, artificial intelligence, etc.
Especially in areas such as 5G communication and artificial intelligence, iterative upgrades are frequent and there is great uncertainty in technology.
Therefore, the market increment of FPGA chips is mainly in 5G base stations and AI artificial intelligence.
In the field of 5G, the amount of FPGA chips for 5G single base stations is expected to increase from 2-3 chips in the 4G period to 4-5 chips in the 5G period, doubling the usage.
In addition, due to the rapid attenuation of 5G signals, the number of 5G base stations has increased significantly compared with the 4G period.
In the field of AI, thanks to the advantages of FPGA chip parallelism and low latency, FPGA is mainly used as an acceleration card in data center computing, and cooperates with CPU in servers.
With the acceleration of the iteration speed of deep learning of artificial intelligence and AI large models, the amount of FPGA chips will open up incremental space.
How big is the market space?
Data shows that the global market size of FPGA chips has increased from 45 in 2013$6.3 billion, up to $63.3 in 2018$3.5 billion. The compound annual growth rate is about 88%。
With the progress of the current 5G era and the advancement of AI, the FPGA chip market space is growing rapidly.
The global FPGA market size is expected to grow from 68 in 2021$600 million to $125 by 2025800 million US dollars, with an average annual compound growth rate of about 164%。
Driven by the demand for domestic substitution, China's FPGA market is growing faster than the global average.
According to statistics, the size of China's FPGA chip market is expected to increase from 39 in 2018700 million yuan, an increase to 459 in 2023400 million yuan, with an average annual compound growth rate of 269%。
It is expected to reach about 79 billion yuan by 2025, with an average annual compound growth rate of 34 from 2023 to 20253%。
Investment opportunities are concentrated in **?
The FPGA chip market space is huge, the market is growing steadily, and there are very few domestic participating companies in such a good industry track. Only Fudan Microelectronics has done better.
As the saying goes, "scarcity is expensive", and without opponents, gross profits are naturally high.
According to the financial report, the gross profit margin of Fudan Microelectronics' FPGA chips is as high as 9526%！Comparable to Moutai, it is estimated that there is only one "manufacturing industry" with such a high gross profit in all A-shares!
The main reason is that FPGA chips are mainly used in specific fields such as high reliability, with extremely high technical thresholds and single application fields, so the gross profit margin is high.
At present, in the context of the continuous iterative upgrading of artificial intelligence, the sales revenue of Fudan Microelectronics FPGA chips is expanding dramatically.
This can be confirmed by the financial data, because after 2020, Fudan Microelectronics is like a different company.
From 2018 to 2020, revenue growth was slow, with an average annual compound growth rate of only 9%.
After 2020, net profit, revenue, and gross profit margin will begin to explode!
First of all, the performance in 2020 turned losses into profits, with a net profit of 13.3 billion yuan, and then a net profit of 5 in 20211.4 billion yuan, a year-on-year increase of 28720%, 10 direct in 20227.7 billion, net profit increased nearly 10 times in two years.
The main reason is that the "high gross margin FPGA chips" began to increase in volume, which made the company's performance grow rapidly.
Among them, the FPGA chip business will achieve operating income of 1 in 20205.3 billion, a year-on-year increase of 8214%。
With a high gross profit of 95% for FPGA chips, the gross profit margin began to enter an upward range after 2020, and the gross profit margin reached 66 in the first quarter of 202377%, an all-time high.
Of course, the revenue has not fallen, and the company will achieve an operating income of 35 in 20223.9 billion yuan, a year-on-year increase of 3731%, and the net profit attributable to the parent company was 107.7 billion yuan, a year-on-year increase of 10931%。
The reason why revenue has doubled in 2 years and net profit has increased 10 times in 2 years is that the downstream artificial intelligence of FPGA chips has exploded.
This is still the data after 2020, in 2023, when the graphics card** will increase fivefold and the inventory will be "snapped up", the revenue in 2023 and 2024 will be more explosive.
In general, the current rate of performance improvement of integrated circuits can no longer match the growth rate of computing performance demand.
Before breakthroughs are made in basic technologies such as chip materials, the use of dedicated "CPU + FPGA accelerator card" is one of the mainstream solutions to improve hardware performance.
Thanks to the unique architecture, FPGA has incomparable advantages in terms of flexibility such as ASIC and GPU, and also has certain advantages in the time to market and cost of superimposed FPGA chips.
Therefore, in the context of the rapid iteration of artificial intelligence, FPGA chips will usher in an explosion, and related products may benefit.
Pay attention to Little Elephant and get the most valuable "Growth 20: 20 Companies with the Most Growth Potential in 2024" investment report now!