Huang Renxun's latest announcement!
On November 13th, Nvidia CEO Renxun Huang announced at the Nvidia Japan Summit that he will collaborate with SoftBank to build AI infrastructure in Japan, including the country's largest AI factory. SoftBank is investing heavily in data center construction in Japan, pressing the 'restart button' for Japan's technological development. Japan cannot miss this round of AI development, we need infrastructure, "said Masayoshi Son, former chairman and president of SoftBank
In addition, according to the latest news, Japan's newly appointed Prime Minister Shigeru Ishiba has promised to provide 10 trillion yen (approximately 470 billion yuan) in new support for Japan's semiconductor and AI industries, in order to keep up with the global investment boom in cutting-edge technologies.
According to a draft plan, the new funding framework will be separated from the previous special fund worth approximately 4 trillion yen and will be included in the upcoming stimulus plan aimed at generating approximately 160 trillion yen in economic benefits.
However, in the capital market, affected by the collective adjustment of US chip stocks, chip stocks in the Japanese stock market also experienced a pullback on Tuesday. Lasertec fell more than 4%, while Edwan Testing and Tokyo Electronics fell 3.2% and 2.8% respectively. Today (Wednesday), the trend of chip stocks in the Japanese stock market was divided. As of press time, Tokyo Electronics rose more than 3%, Renesas Electronics fell more than 2%, Lasertec fell 0.30%, and Edwan Test rose 0.20%.
NVIDIA collaborates with SoftBank
On November 13th, Nvidia CEO Renxun Huang announced at the Nvidia Japan Summit that he will collaborate with SoftBank to build AI infrastructure in Japan, including the country's largest AI factory.
SoftBank Group's founder Masayoshi Son said, "SoftBank is investing heavily in data center construction in Japan, pressing the 'restart button' for Japan's technological development. Japan cannot miss this round of AI development, we need infrastructure
Huang Renxun stated that SoftBank is using the NVIDIA Blackwell platform to build Japan's most powerful artificial intelligence supercomputer and plans to use the NVIDIA Grace Blackwell platform for its next supercomputer.
In addition, Nvidia revealed that SoftBank has successfully tested the world's first combination of artificial intelligence and 5G telecommunications networks using Nvidia's AI Aerial acceleration computing platform, which is a breakthrough in the field of computing and opens up potential AI revenue streams worth billions of dollars for telecommunications operators.
Nvidia and SoftBank also announced that SoftBank's goal is to create an AI marketplace using Nvidia AI Enterprise software that can meet local security AI computing needs. This new service that supports AI training and edge AI inference positions SoftBank as Japan's AI grid, creating new business opportunities for various industries, consumers, and businesses across Japan to create, distribute, and use AI services.
Junichi Miyakawa, President and CEO of SoftBank, said, "Countries and regions around the world are accelerating the adoption of artificial intelligence to promote social and economic growth, and society is undergoing significant transformation. Through our long-term partnership with NVIDIA, SoftBank is leading this transformation. With our extremely strong AI infrastructure and our new distributed AI-RAN solution 'AITRAS' that reshapes 5G networks for AI, we will accelerate innovation nationwide and around the world
It is reported that SoftBank will acquire the world's first NVIDIA DGXB200 system, which will serve as the building block for its new NVIDIA DGX SuperPOD supercomputer. SoftBank plans to use its Blackwell based DGX SuperPOD for its own generative artificial intelligence development and AI related businesses, as well as for universities, research institutions, and businesses across Japan.
In addition to DGX SuperPOD, SoftBank also plans to build another NVIDIA accelerated supercomputer to run extremely computationally intensive workloads. The preliminary plan for this supercomputer is based on the NVIDIA Grace Blackwell platform design, which adopts the NVIDIA GB200 NVL72 multi node, liquid cooled, rack level system, combining NVIDIA Blackwell GPU with Arm based energy-saving NVIDIA GraceCPU.
Japan invests heavily
On the evening of November 11th local time, Japan's newly appointed Prime Minister Shigeru Ishiba announced an ambitious investment plan. Shigeru Ishiba proposed that the Japanese government will provide at least 10 trillion yen in funding support before the 2030 fiscal year to promote the semiconductor and artificial intelligence industries.
At the press conference that evening, Shigeru Ishiba stated, "We will develop a new aid framework to attract over 50 trillion yen in public and private investment over the next 10 years." This plan will be included in the comprehensive economic plan determined in November, providing subsidies, investments from government affiliated institutions, and loan and debt guarantees for private sector financial groups. Shigeru Ishiba emphasized that the government will not issue "deficit making bonds" to fund the initiative.
According to a package plan draft obtained by the media, the new funding framework will be separated from the previous special funds of about 4 trillion yen and will be outlined in the upcoming economic stimulus plan, with the goal of generating an economic impact of about 160 trillion yen.
The special fund of 4 trillion yen was proposed by the cabinet led by former Japanese Prime Minister Fumio Kishida. This budget aims to revitalize the chip industry, including allocating 920 billion yen to local chip company Rapidus, whose goal is to mass produce advanced logic chips by 2027. The newly announced 10 trillion yen additional funds have a broader scope of use, aimed at helping Japan narrow the gap with global chip support powers.
At present, countries are competing to invest more in semiconductor capabilities driven by artificial intelligence, and policy makers now believe that this field is crucial for economic security. For example, US President Biden's 2022 Chip and Science Act promises to provide chip manufacturers with a total of $39 billion in grants, as well as $75 billion worth of loans and guarantees, plus up to 25% tax credits.
Based on the need to promote economic growth, the Japanese government is also increasing its support for this industry. According to the proposed plan, global chip demand is expected to double in the next 10 years, reaching 150 trillion yen. This framework aims to provide over 10 trillion yen in public assistance through outsourcing, financial support, and legislative measures to enhance the predictability of private companies. From the perspective of economic security, the Japanese government believes that it is necessary to establish the most advanced technology in the semiconductor field. If subsidies are gradually invested on a single year basis, the predictability is low, so it will be changed to planned support for multiple years.
Japanese Minister of Economy, Trade and Industry Yoji Muto stated on Tuesday that the Japanese government will not increase taxes to fund the new framework, and added that details are still being finalized. Shigeru Ishiba pointed out that he will discuss the financing issues of the plan with various departments, but he will not use deficit financing bonds as funds to pay for these measures.
SoftBank accelerates AI layout
On November 12th, SoftBank released its performance report for the second quarter (third quarter of the calendar year) and the first half of the 2024 fiscal year ending at the end of September. Data shows that SoftBank's revenue in the first half of the year increased by 7% year-on-year to 315 trillion yen, achieving a net profit of 1.18 trillion yen, far exceeding market expectations of 295 billion yen. SoftBank's Vision Fund turned losses into profits in the second quarter, with a profit of 373 billion yen, mainly driven by the IPO boom in India.
Currently, Masayoshi Son, the founder of SoftBank Group, is aggressively entering the fields of artificial intelligence and semiconductor investment. Sun Zhengyi is raising funds to enter the field of artificial intelligence and related hardware by selling assets in the Vision Fund investment portfolio. In May of this year, the company announced an annual investment plan of $9 billion in artificial intelligence.
Although Masayoshi Son has hinted at the possibility of launching a new Vision Fund series, the possibility of launching a third or fourth fund no longer exists. On the contrary, SoftBank's current investments are mostly carefully planned by holding companies and bypass the Vision Fund. Son's new pursuit is partly inspired by the success of Arm, whose market value has skyrocketed to around $150 billion since going public last year, with SoftBank holding 90% of the company's shares.
Recently, SoftBank contributed $500 million to OpenAI's funding round. Earlier in June, Softbank Group's Vision Fund announced on the 2nd that it would invest in Perplexity AI, an American AI startup. In addition, Graphcore, a chip design company acquired by SoftBank Group, is adding 75 new jobs globally, with a 20% increase in employee numbers, highlighting SoftBank's ambitious goals in the semiconductor industry. Graphcore announced new positions covering silicon wafer design and data center engineering.
At the end of October, Masayoshi Son stated in a media interview that the sustained spending trend in the field of artificial intelligence will only accelerate in the coming years, which will benefit Nvidia and its GPU chip business, which has been the driving force behind most advances in artificial intelligence. I think Nvidia is underestimated because the future of (artificial intelligence) is much broader, "said Masayoshi Son at the Saudi Future Investment Initiative Summit. Sun Zhengyi is optimistic about artificial intelligence because he believes that "super artificial intelligence" will be achieved by 2035.
Artificial intelligence is widely regarded as an AI system that is as intelligent as humans, and Son defines super AI as an AI system that is 10000 times more powerful than the human brain. Sun Zhengyi stated that to achieve this level of artificial intelligence, companies will need over 200 million GPU chips and invest a total capital expenditure of $9 trillion.
Sun Zhengyi stated that considering the potential huge profits that super artificial intelligence could bring, this could be a cost-effective deal. I think this is still a very reasonable capital expenditure. $9 trillion is not too much, maybe too little, "he said. Sun Zhengyi also stated that if super artificial intelligence replaces 5% of GDP within 10 years, it could potentially bring in profits of up to $4 trillion annually.