Image source @ Visual China
Recently, Huawei founder Ren Zhengfei said in an article that the global economy will face a recession and a decline in spending power, 2023 and 2024 is Huawei's breathing period, and the entire company will take "survival" as the main program in the future. "The world's economy is unlikely to improve in the next 3 to 5 years."
Ren Zhengfei's "cold gas theory" has aroused heated discussion on the Internet in Chinese. At the same time, this "cold gas" also spread to the "chip big three" in the United States - Intel, AMD, and NVIDIA.
In late August, U.S. chip giant Nvidia (NASDAQ: NVDA) released its financial report for the second quarter of fiscal 2023 as of July 31 this year, showing that Nvidia's revenue for the quarter reached $6.704 billion, an increase of only 3% year-on-year; Net profit was $1,292 million, down 51% year-over-year. Among them, the once speculated on the sky-high price gpu (graphics processor) graphics card is the volume and price of the fall, the long-term contribution of Nvidia nearly half of the revenue of the game business, down 33% year-on-year.
Coincidentally. Intel (NASDAQ: INTC) and AMD Corporation (NASDAQ: AMD) have also reported their latest quarterly results in the past month. Among them, Intel's revenue in the second quarter plunged by 22% year-on-year, non-GAAP revenue fell by 17% year-on-year, and net profit turned from profit to loss; AMD also predicted signs of slowing revenue growth in the third quarter.
When the cold hit, how the US chip giants solve short-term interests and long-term development seems to have become an important topic that we need to pay attention to at present.
3 chip stocks are polarized: Intel loses share, AMD is undervalued
Fears of a global recession have exacerbated forecasts of declining global chip sales.
The World Semiconductor Trade Statistics Organization (WSTS), a nonprofit that tracks chip shipments, predicts that this year's chip market growth forecast will be reduced from 16.3% to 13.9%, but the trade size will still exceed $600 billion. At the same time, chip sales in 2023 will only grow by 4.6%, and the growth rate will hit the lowest since 2019.
At present, there are signs of recession in the world, chip competition in various countries has intensified, fierce inflation has eroded the purchasing power of consumers, sales of smart phones and PCs have also slowed down, the demand for CPU (central processing unit), GPU (graphics processor) chips has decreased, and the rapid interest rate hikes in many economies are threatening global economic growth, which has also made enterprises more cautious about capital expenditure. This is the fundamental reason for the slowdown in the performance growth of the three chip companies and the decline in profits.
Specifically, the performance of these three chip giants in the second quarter of each has a very different development.
A data released by Ycharts shows signs of polarization in the annual revenue growth trend of the three chip stocks - Nvidia and AMD are growing at a rate of more than 140%; Intel, on the other hand, slowed its growth rate by 9.81%, lost market share, lowered its guidance of expectations, and became the company with the lowest growth and market capitalization among the three companies, lower than Wall Street's expectations.
Among them, Intel's financial performance in the second quarter was lower than market expectations, and revenue and profit fell twice: revenue in the second quarter was $15.321 billion, down 22% year-on-year; Net loss was $454 million, compared to net profit of $5,061 million in the year-ago quarter, down 109% year-over-year; Diluted loss per share was $0.11, down 109% year-over-year.
According to market research firm IDC data, in the second quarter of 2022, global PC shipments fell by 15.3% year-on-year, totaling 71.3 million units. The PC chip business has long contributed more than half of Intel's revenue. In the second quarter of 2022, Intel's PC chip business revenue was $7.7 billion, down 25% year-over-year.
AMD (Ultra Micro Semiconductor) revenue in the second quarter was slightly better, but the profit was still very low. AmD's revenue in the second quarter reached $6.6 billion, up 70% year-on-year, thanks to the merger of xilinx's results and other factors; Net profit was $447 million, down 37% year-over-year; Earnings per share were $0.27, down 53% year-over-year. But AMD expects Q3 revenue to be around $6.7 billion, below analyst average expectations.
NVIDIA's revenue growth was slightly better than THAT of AMD and the most expensive of the three chip stocks. This is mainly due to the growth in the demand for computing power, games, crypto mining, AI and cloud computing and other fields are very much in need of GPU performance changes, so that NVIDIA has become a capital market "fragrant", the market value is at a fairly high level - second only to TSMC, the world's second largest chip stock.
According to the financial report, in the fiscal year 2022 ended January 30, 2022, Nvidia's full-year revenue hit a record of $26.91 billion, an increase of 61% year-on-year; Non-GAAP diluted earnings per share for the full year were $4.44, up 78% year-over-year.
However, analyst Jonathan Weber said in an article that NVIDIA's valuation can be said to be ridiculously high, the stock price is 50 times the expected net profit this year, the price-earnings ratio is as high as 31 times, compared to AMD, NVIDIA's transaction price premium of more than 100%, is not reasonable. He believes that the next NVIDIA stock price will enter a period of decline, while AMD is undervalued, there will be rapid growth in the future, the current price-earnings ratio is about 18 times.
In terms of performance, AMD's revenue in the second quarter increased by more than 70% year-on-year, and it is expected that revenue will reach $12.4 billion in the first half of this year, and full-year revenue is expected to reach $26.3 billion.
This means that compared with Nvidia and Intel, AMD's revenue in the second half of the year should be around $14 billion, an increase of more than 10% month-on-month, while the other two will not see double-digit growth in the second half of the year. However, this premise is that AMD will not be affected by the epidemic and market changes in the second half of the year. Because Nvidia has predicted that its third-quarter revenue will be about $5.9 billion, which is lower than market expectations, and said sales will fall 17% year-on-year.
Changes in market capitalization (share price) of Nvidia and AMD
On August 31, Jon Peddie Research released data showing that in the second quarter of 2022, independent and integrated GPU sales fell by 14.9% sequentially. Among them, Nvidia GPU sales fell by 25.7% month-on-month, AMD fell by 7.6%, Intel fell by 9.8%, and sales are expected to continue to decline in the third quarter.
The cold is coming, and the big three save themselves
As we all know, chip semiconductors are a strong cyclical track, which requires long-term research and development investment and a global supply chain system.
However, investors in the capital markets are more focused on the short-term interests of businesses. For example, while AMD presented investors with a total of $300 billion in market opportunities over the next five years, including data centers, gaming, smart driving, and more, stocks did not see a big rally.
When the consumer electronics market is weak, cloud and data center growth is slowing, and the short-term "cold" is coming, the "chip big three" in the United States, NVIDIA, AMD, and Intel, have begun self-help work: continuous activity and complementary cooperation between companies.
On August 30, AMD launched the AMD Ryzen 7000 series CPU chips in the United States, based on TSMC 5nm process nodes, using Zen 4 architecture, compared with the previous generation Zen 3 to bring 13% IPC (interprocess communication) improvement, single-threaded performance increased by 29%, multi-threaded performance increased by 35%.
Compared with Intel's 13th generation Core series CPU, AMD's new model release time is much earlier, and it has become the first 5nm CPU chip product. At the press conference, AMD unveiled the Zen architecture roadmap. According to the roadmap, in 2024, AMD will release a Zen 5 architecture processor chip using the 4nm/3nm process.
Although no new products have been launched, Intel is not idle. On August 23, local time, the US chip giant Intel announced that it has signed an agreement with Canadian asset management company Brookfield to jointly provide up to $30 billion in investment for Intel's two new fabs, and the transaction is expected to be completed by the end of 2022. Intel will invest 51 percent and Brookfield will invest 49 percent, and Intel will own and operate two fabs.
At the same time, since the Chip and Science Act signed by US President Biden took effect last Thursday (25th), the bill includes subsidies of $52.7 billion for the chip industry, a 25% investment tax credit for semiconductor and equipment manufacturing, and supportive policies such as investment in wireless broadband technology. Intel CFO David Zinsner said Intel is continually seeking to work with local and U.S. federal governments.
According to previous US technology media reports, Intel is lobbying the Biden administration to get up to one-third of the financial subsidy support from $52.7 billion, which is estimated to be about $17.6 billion.
According to a roadmap obtained by titanium media app, in 2023, Intel GPU products will be produced using TSMC 5nm (N5) process, SoC Tile and IOE Tile will be produced using TSMC 6nm (N6) technology nodes, and CPU will be Intel 4 nodes. Therefore, this means that intel's factories in the United States will not produce the most advanced processes in the short term, and it is expected that up to 14nm processes, advanced chips will be transferred to TSMC production.
In addition, in order to save their own performance, companies have also carried out complementary cooperation.
Among them, NVIDIA and Ampere Computing jointly announced the launch of an AICAN cloud server technology platform. Earlier, Intel announced a chip foundry cooperation with MediaTek; Qualcomm announced a cooperation with Samsung to install the Qualcomm Snapdragon mobile chip on Samsung's Galaxy mobile phone.
On the whole, with the development of data centers, PC games, and the continuous expansion of computing power demand, the chip semiconductor market will have a large growth trend in the long run. In the CPU and GPU markets alone, AMD and Intel's GPU market share is slightly lower than that of Nvidia, while Intel still dominates the data center server CPU, but the growth rate is slowing, and AMD is eroding its share.
Taking cars as an example, tesla model S and Model X are equipped with AMD Ryzen embedded APUs and RDNA-2 GPUs respectively, while Intel's mobileye self-driving products have also been seized market share by NVIDIA Orin.
Wang Rui, senior vice president of Intel and chairman of Intel China, told Titanium Media App last year that although they have lagged behind in the process in the past few years, they have actively caught up, hoping to surpass AMD and Nvidia and regain the favor of customers and the advantages of the market. "Intel is confident in winning back customers like Apple."
So, what are the implications for China from the development and changes of the "chip big three" in the United States?
In fact, as a popular track for R&D in the chip industry, China is constantly increasing the research and development of next-generation computing chip products such as CPUs, GPUs, FPGAs, etc., in order to become "China NVIDIA", "China Xilinx", "China Intel", including Tianshu Zhixin, Flinty Technology, Biling Technology and other enterprises have gradually achieved commercialization, of which the valuation of Biyi is as high as 15 billion yuan.
If the news that the United States has cut off the supply of high-performance GPU chips to China is true, this will benefit the entire domestic CPU/GPU chip design enterprises and the large chip industry, and the Americans will force Chinese customers to use domestic GPUs.
However, challenges remain. Since such chip tracks began to chase advanced process processes, and the highest process Chinese mainland was 14nm, the CPU/GPU has long relied on TSMC's process. In addition, the current proportion of CPU computing resources is very low, only less than 0.1%; The resource utilization utilization rate of FPGA hardware is less than 5%; The energy consumption of GPU data transmission is as high as more than 90%. Existing computing chips are not up to the task of next-generation computing.
"If you want to complete the super operation, the energy consumption is 8,000 megawatts, and Beijing is 25,000 megawatts for the whole year of 2021, it will use 1/3 of Beijing's electricity, costing 4 trillion yuan, then China's annual GDP is 115 trillion yuan, and it will use about 2% to 3% of the country's GDP to build a computer, which is very costly." Therefore, it is very difficult for the world to conquer the next generation of computers. Wei Shaojun, a professor at Tsinghua University, said at an event on the 30th that the current 14nm and international advanced 3nm processes have a performance gap of about 10 times, which is a major challenge for us.
Wei Shaojun stressed that in the process of development, China should re-examine this year's chip architecture, and chip architecture innovation will become one of the important tasks. China's integrated circuit design industry has achieved certain results, with a considerable industrial scale, and the next stage from catching up to running, and gradually in some areas to achieve such a development process.
But now, The R & D investment of Chinese chip companies is seriously insufficient, "we must have government support for innovation, we must have talents, if there is no investment, it is still impossible to do ... Therefore, in the matter of R&D investment, it is not only the responsibility of the central government, nor is it just the responsibility of entrepreneurs, and I hope that local governments can also work together to promote it. Wei Shaojun called for China to increase support for semiconductor companies' R&D investment at this year's Wuxi ICDIA conference. (This article was first published on titanium media App, author | Lin Zhijia)