Big data and the Internet have joined hands to usher in a new era of interaction. Companies that once benefited from the Internet wave have also ushered in a new round of transformation. Competition around artificial intelligence and chips has become the key "lifeline" in this round of transformation. Therefore, many Internet giants have begun to get involved in the semiconductor industry.
As a newcomer in the Internet industry, ByteDance has risen rapidly under the dividends of extensive reading, and then stepped on the pulse of the times when Internet companies entered the semiconductor industry. Under this arrangement of fate, ByteDance also began to try to make a big splash in the chip field, making the "bytes" of the Internet "beat" more powerfully.
ByteDance’s chip chess game
According to people familiar with the matter, ByteDance is developing its own cloud AI chips and Arm server chips.
According to previous public information in the market, ByteDance is actively building an AI chip team, and currently has many chip-related positions on major recruitment platforms. According to information from insiders, the research and development direction of ByteDance's AI chips has been clarified.
ByteDance chose to enter the chip field from cloud AI chips and Arm server chips, which is also the mainstream way for Internet companies to enter the chip market. From the perspective of the international market, since Google released TPU in 2016, it has become a trend for international Internet giants to enter the chip field. Later, domestic BAT Internet giants also expanded their businesses, and Internet companies have become unstoppable in making chips. As the first manufacturers to take the lead, these Internet giants are deploying cloud AI chips or Arm server chips.
As for the AI cloud chip market, according to the latest survey report of ABI Research, the global cloud AI chip market is expected to reach 10 billion US dollars in 2024. However, judging from the current market situation, NVIDIA has an absolute advantage in the AI cloud chip market. However, for different Internet companies, due to the different ecosystems they have built, their requirements for chip performance are different. In this case, customized chips may enable them to better play the value of the ecosystem. Judging from the current market situation, the market has not given Internet companies enough choices. In this case, self-developed chips have become a development path.
However, compared with AI cloud chips, which are more difficult to develop and more targeted, GPUs based on AI acceleration are currently more mainstream in the cloud. Therefore, ByteDance has also deployed in the GPU field - in February this year, ByteDance invested in Moore Thread, a local GPU chip design rookie.
On the other hand, considering ByteDance's business, both its Toutiao and Douyin platforms will carry more content in the 5G era, which requires the support of server chips such as data centers and edge computing. At the same time, as the market demand for server chips increases, open source architectures such as Arm and RISC-V begin to enter the server market. Many manufacturers also want to take this opportunity to enter the server market originally occupied by X86, and Internet companies are one of them.
As for the Arm server chip market, this emerging semiconductor market has attracted many manufacturers to participate, and Marvell is one of the best. However, after several years of exploration, Marvell began to shift from the general market to the customized market. This also shows that a specific market environment may be more conducive to the commercialization of Arm server chips. As users of server chips, Internet companies are more aware of the chip performance required by their ecosystem. The rise of Arm server chips has put them on the same starting line as semiconductor manufacturers. In addition, their own understanding of demand may help them become dark horses in the chip field through the server chip market.
For Internet companies, joining the competition for server chips can reduce their dependence on third-party suppliers and reduce costs by developing their own chips. This can be proven by the international Internet giant Amazon. According to previous reports by foreign media, after launching two generations of data center chips, Amazon's self-developed chips have also begun to be used to process some computing tasks of the Alexa voice assistant. In early tests, the processing results of Amazon's self-developed chip Inferentia cluster were the same as those of Nvidia's T4 chip, but the latency was reduced by 25% and the cost was reduced by 30%.
Considering ByteDance's situation, choosing to develop the server chip field based on the Arm architecture is not only in line with the current market trend, but also beneficial to the company's long-term development.
In addition to the Arm architecture being sought after by many manufacturers, RISC-V is also considered by the industry to be another open source architecture that can realize the implementation of artificial intelligence technology. Therefore, in January this year, ByteDance also invested in Shim Computing, which is committed to developing a dedicated architecture processor for the field of artificial intelligence based on the RISC-V instruction set architecture. According to its official website, the NeuralScale NPC core architecture independently developed by Shim Computing is a world-leading, dedicated computing core for the field of neural networks that is expanded based on the RISC-V instruction set. It has world-leading power efficiency and extreme programmability, and can meet the needs of diverse artificial intelligence algorithms and applications in the cloud.
Behind ByteDance’s chip-making
This is an era where heroes create the times, and the times create heroes. The rise of foundries such as TSMC has created a new era for the semiconductor industry. Their arrival has lowered the threshold for companies to participate in the chip field. Fabless manufacturers can save costs in chip manufacturing, allowing them to focus more energy and capital on chip design. Under this trend, Fabless manufacturers have been able to grow rapidly, driving the rapid development of the semiconductor industry, and also opening the door to the semiconductor field for new players such as ByteDance.
In addition to the convenience of chip manufacturing, the open chip architecture licensing model has also brought development opportunities to chip design manufacturers. The rise of emerging markets such as artificial intelligence has caused turmoil in the semiconductor industry, which has opened up opportunities for all players to enter the semiconductor industry. Open source provides all new players with a sword that can cut through thorns and brambles. Open source architectures such as ARM and RISC-V allow more players to have the opportunity to participate in this world called chip competition. The emergence of these open source architectures has made chip design no longer an industry with high exclusivity and difficult barriers to cross. This licensing model has been recognized by many semiconductor manufacturers and has also provided a development path for Internet players involved in the semiconductor field.
From another perspective, Internet companies previously used third parties to supply the chips they needed, but as far as the current market situation is concerned, shortages will appear in the market from time to time. For example, GPUs are currently out of stock, which may affect the supply of the cloud market. Therefore, ByteDance's self-developed chips may allow them to take more initiative when facing shortages.
In addition, a series of bans issued by the United States on the development of China's semiconductor industry have also greatly promoted the prosperity of China's local semiconductor companies. This may also have given ByteDance some inspiration, allowing them to participate in the localization track.
In our view, these external forces have not only pushed ByteDance into the chip field, but also hundreds of manufacturers involved in the chip field as a result will eventually merge into an ocean like small streams, thus affecting changes in the semiconductor industry.
ByteDance's Sea of Stars
Judging from the chip business currently deployed by ByteDance, they have mainly deployed a large number of cloud chips. In addition to the AI cloud chips mentioned above, Toutiao, a subsidiary of ByteDance, is also actively deploying in the cloud computing market.
The 2020 epidemic has accelerated the digital transformation of various enterprises, and the enterprise service market, especially the enterprise technology service market, has experienced an explosion under this background. This situation has also opened the door to To B for ByteDance, and ByteDance Volcano Engine is the key to open this door.
The cloud market has not only attracted the attention of ByteDance, but also many Internet companies. Compared with those Internet companies that were established earlier, their ambitions may not only be in the cloud market.
Take Amazon as an example. Amazon's chip business has already involved Arm server chip Graviton and machine learning chip Inferentia. At the same time, they have also invested in some chip design manufacturers. Their business not only involves the cloud, but also the edge and terminal are also important markets in the era of artificial intelligence.
The market believes that future computing will not only be limited to large data centers, but will be distributed on an integrated continuous spectrum consisting of cloud, edge, and end. Cloud and edge are also the core factors driving the popularization of the Internet of Things. Various smart terminals use the Internet of Things cloud platform to achieve interconnection. Therefore, expanding into other markets besides the cloud market may allow ByteDance's chip business to go further.
This era has given ByteDance the opportunity to participate in the chip competition, but what impact ByteDance will have on the development of the semiconductor industry in the future remains an unknown.
Previous article:Former TI Vice President Hu Yuhua officially becomes President of Goodix Technology
Next article:Synopsys and Xinyaohui join forces to accelerate the pace of domestic chip innovation
- Popular Resources
- Popular amplifiers
- Vietnam's chip packaging and testing business is growing, and supply-side fragmentation is splitting the market
- The US asked TSMC to restrict the export of high-end chips, and the Ministry of Commerce responded
- ASML predicts that its revenue in 2030 will exceed 457 billion yuan! Gross profit margin 56-60%
- ASML provides update on market opportunities at 2024 Investor Day
- It is reported that memory manufacturers are considering using flux-free bonding for HBM4 to further reduce the gap between layers
- Intel China officially releases 2023-2024 Corporate Social Responsibility Report
- Mouser Electronics and Analog Devices Launch New E-Book
- AMD launches second-generation Versal Premium series: FPGA industry's first to support CXL 3.1 and PCIe Gen 6
- SEMI: Global silicon wafer shipment area increased by 6.8% year-on-year and 5.9% month-on-month in 2024Q3
- LED chemical incompatibility test to see which chemicals LEDs can be used with
- Application of ARM9 hardware coprocessor on WinCE embedded motherboard
- What are the key points for selecting rotor flowmeter?
- LM317 high power charger circuit
- A brief analysis of Embest's application and development of embedded medical devices
- Single-phase RC protection circuit
- stm32 PVD programmable voltage monitor
- Introduction and measurement of edge trigger and level trigger of 51 single chip microcomputer
- Improved design of Linux system software shell protection technology
- What to do if the ABB robot protection device stops
- Learn ARM development(15)
- Analysis of the application of several common contact parts in high-voltage connectors of new energy vehicles
- Wiring harness durability test and contact voltage drop test method
- From probes to power supplies, Tektronix is leading the way in comprehensive innovation in power electronics testing
- From probes to power supplies, Tektronix is leading the way in comprehensive innovation in power electronics testing
- Sn-doped CuO nanostructure-based ethanol gas sensor for real-time drunk driving detection in vehicles
- Design considerations for automotive battery wiring harness
- Do you know all the various motors commonly used in automotive electronics?
- What are the functions of the Internet of Vehicles? What are the uses and benefits of the Internet of Vehicles?
- Power Inverter - A critical safety system for electric vehicles
- Working principle of lithium battery equalization circuit
- Have you fallen into these customer support "pitfalls"? Read the story, write a comment, grab a post and win a gift!
- About rail-to-rail op amps
- 【TGF4042 signal generator】+ Load capacity test
- EEWORLD University - In-depth understanding of PCI Express 5.0 testing
- LTspice .subckt(3) reverse drawing
- TI blog post: Buck-boost DCDC TPS63810 in TWS headphones
- USB5537B, does not start after first power on, but restarts normally after power off
- Showing off the EEWorld 15th Anniversary T-shirt
- Dating Spring - Vegetable Garden and Fruits at Home