Tech giants are betting heavily on the AI war

China Business News

2024-05-21 07:51Published on the official account of China Business News in Beijing

Reporter Wu Qing reports from Beijing

AI (Artificial Intelligence) has become a keyword in the global technology venture capital community in the past year, and it is likely to only get more lively in the future.

In the past week, the global tech giants have intensively released the latest AI products and technologies. On May 13, OpenAI used GPT-4o to steal attention. A day later, Google also released the AI product Family Bucket, and the Veo video model, Project Astra, and the new version of AI search all left an amazing memory point. Multimodal fusion technology has become the focus of both of them, both of which are moving towards the goal of AGI (Artificial General Intelligence).

On the other side of the ocean, ByteDance released the bean bag model family, and Tencent finally handed over the answer sheet of the intelligent twink product "Tencent Yuanqi" and the large model assistant APP. Kai-Fu Lee, who has zero one thousand things, officially unveiled the 100 billion parameter Yi-Large closed-source model, and also disclosed the strategic layout of the open-source closed-source dual-track model.

In the near future, there will also be a blockbuster conference of two technology giants: on May 21, Microsoft will hold the Build annual developer conference; On June 10, Apple will host its Worldwide Developers Conference.

A number of industry insiders said in an interview with a reporter from China Business News that the rapid detonation of the artificial intelligence track is becoming the focus of global attention, and the AI competition of technology giants will continue to heat up in the future. Driven by the strong demand for AI, major manufacturers are concentrating their firepower to accelerate their layout. At present, Google, Microsoft, Amazon, Apple, Meta and other technology companies have begun to bet heavily on the field of artificial intelligence, and continue to raise capital expenditures.

Tech giants pull together

After the release of ChatGPT, a chatbot model, and Sora, a large Wensheng video model, OpenAI recently launched its latest achievement, GPT-4o.

The release of GPT-4o has sparked widespread discussion in the AI community. The industry believes that the amazing thing about GPT-4o lies in two points: first, it shortens the delay of voice interaction to 300ms; The second is the end-to-end multimodal native large model. The voice assistant has also taken a qualitative leap in comprehension, not only analyzing and discussing the images or videos it sees, but also recognizing the different emotions of the user when speaking.

Just a day later, Google unveiled AI products at the I/O 2024 developer conference, from search to Gmail, TPU, to voice assistant Astra and multimodal video model Veo. In the 110-minute speech, Google CEO Pichai mentioned AI 121 times.

At the same time, on the other side of the ocean, in China, ByteDance's self-developed large model bean bag large model (formerly the lark large model) family made its debut with 9 models. The inference price of the bean bag model has become a highlight, and its main model is priced at only 0.0008 yuan/1000 Tokens in the enterprise market, and 0.8 centimeters can process more than 1,500 Chinese characters.

On May 17, Tencent's hybrid Wensheng diagram model was announced to be open source, which has been released on the Hugging Face platform and Github, including complete models such as model weights, inference codes, and model algorithms, which can be used for free commercial use by enterprises and individual developers. Tencent also announced the upgrade of the hybrid model, launching three model versions with different characteristics in terms of quality and cost, and it already has more than 600 business access models.

After nearly a year of continuous "wild running", 010000 Things has also officially released Yi-Large with a parameter scale of 100 billion yuan recently. Kai-Fu Lee also announced the global launch of the "API Open Platform" that empowers the 2B ecosystem.

In the coming period, AI will continue to be the dominant character.

More than half of the total number of sessions related to "AI" at Microsoft's annual developer conference, which will be held on May 21. Copilot will be one of the key topics of the Microsoft Build annual developer conference. Microsoft recently announced the integration of AI technology into the development environment.

The Worldwide Developers Conference (WWDC24) will be held from June 10 to 14 local time, and Apple hopes to catch up with its competitors in artificial intelligence. Industry insiders believe that iOS18 will be a "relatively groundbreaking" software update, and artificial intelligence is expected to be integrated into the entire operating system. Some industry insiders predict that Apple may launch a new and upgraded Siri voice assistant, and a new generative AI system has been introduced into the underlying technology.

According to the latest news, Apple may reach a partnership with OpenAI to use external AI in the iOS system, while also in talks with Google about introducing Gemini.

In the short term, look at the market capitalization, and in the long term, look at the application

On the evening of May 14, the two Internet giants Alibaba and Tencent released their financial reports almost at the same time. The financial report shows that the two giants are gradually finding a new round of incremental momentum: artificial intelligence. Through a new round of large-scale investment in artificial intelligence and the support of various businesses, Internet giants expect to gather greater momentum and lead the industry back to the upward path.

According to research reports from a number of institutions, in the medium and long term, the company is actively deploying the development of AI technology, and the diversified product matrix corresponds to the broad direction of potential applications, and technological progress is expected to become an important growth multiplier in the medium and long term.

However, the new wave of AI technology has only just begun, except for a few giants such as Nvidia and AMD, which directly benefit from the "AI era shovel sellers", Internet technology giants are still in the stage of large-scale investment in "burning money".

Tencent revealed that R&D expenditure in the first quarter of 2024 reached 15.6 billion yuan, and R&D investment exceeded 285.3 billion yuan in the six years since 2018. In terms of Alibaba, not only Taotian has set up an AI e-commerce team internally, and the international station has released full-link AI foreign trade products, Alibaba Cloud AI has invested in the product matrix of cloud computing, especially AI infrastructure, in addition to the Tongyi Qianwen model, Ali has also completed the investment in China's five head large model unicorns (Dark Side of the Moon, MiniMax, Zhipu, Baichuan, and Zero One Everything).

Relatively speaking, the American giants are "all in AI".

According to the financial report, Microsoft's capital expenditure in the second quarter of fiscal 2024 was $11.5 billion, another seven-year high. Google's capital expenditures are also expanding rapidly. In the fourth quarter of 2023, Google's capital expenditure was $11 billion, up 45.1% year-over-year, and the company revealed that capital expenditure will continue to increase in 2024 to support continued investment in AI.

Meta and Apple have also expressed their importance and expectations for the future of artificial intelligence business. For example, in order to accelerate the development of AI products, the company plans to raise its capital expenditure in 2024 to $35 billion to $40 billion from the previous forecast of $30 billion to $37 billion.

Compared with the rapidly rising investment, the addition of AI dividends to performance has not yet been clearly reflected in the latest financial reports of the giants. For example, expectations are high for Microsoft's AI business, which has invested in OpenAI. However, Azure, Microsoft's flagship cloud service product in the latest fiscal quarter, continued the growth trend of the previous quarter, but the growth rate was basically the same as the previous quarter after excluding the impact of exchange rates, and did not show the expected high growth.

Lu Yanxia, research director of IDC China, said that the impact of AI on enterprise performance is not significant at present, and the application of large models is not yet mature. Some subdivisions that can be applied will definitely help to improve the performance of enterprises, but the huge cost investment at the beginning will slow down the performance and bring financial pressure to a certain extent.

After the frenzy, will the AI boom cool down in the future? Lu Yanxia believes that it depends on whether large-model AI can be widely applied. If it can't be widely and massively applied, it may cool down or even cool down.

At present, the significance of AI to giants is more the enthusiasm of the capital market and the pull of the stock market.

Relevant data shows that in 2023, AI chip giant Nvidia will rise nearly 5 times, and its market value will climb to $1.8 trillion, becoming the third largest technology company in the world by market capitalization, ARM and AMD will rise 3 times, and Intel will rise by more than 58%; Microsoft also used its investment in ChatGPT to nearly double its stock, surpassing Apple to become the world's most valuable tech giant.

The initiator of this AI boom, OpenAI's valuation has soared, and it has now become the world's third largest technology unicorn after ByteDance and SpaceX.

It is conceivable that artificial intelligence is one of the main lines of investment in 2024, whether from the perspective of policy, industrial trends, or enterprises.

However, an AI industry practitioner told reporters that after the popularity of artificial intelligence stocks last year, investors will become more rational and prudent this year, and the real artificial intelligence winners may be distinguished from those who are "indiscriminate" like C3.ai.

China's AI needs to catch up

Compared with the All in AI of the technology giant on the other side of the ocean, although the domestic war of 100 models has quickly started, there are still some "thunders and rains". Because in terms of investment and financing of AI start-ups last year, only the United States rose, and Europe and Asia both declined.

According to CB Insights, in 2023, the number of AI startups in the United States will reach 1,151, with a total financing of $31 billion, a year-on-year increase of 14%, and it has attracted 46% of all AI transactions worldwide. In 2023, the number of domestic AI investment and financing will be about 232, a year-on-year decrease of 38%; The total amount raised was about $2 billion, down 70% year-on-year. This means that although major domestic giants have launched self-developed large models, the overall AI investment is still declining.

"Fortunately, in the era of AI, China has its own unique talent advantages. On the one hand, the development of China's Internet industry is second only to that of the United States, and China's Internet giants have accumulated talents and technologies for their long-term investment in artificial intelligence. Yang Fei, an observer of the industrial economy, told reporters that on the other hand, the proportion of Chinese among the leading AI companies in the United States is very high, and these people have close ties with the country and will have the opportunity to attract them back to work for the country in the future.

For example, a quarter of the engineers in OpenAI's ChatGPT R&D team are Chinese, and 4 out of 13 people in Sora's R&D team are Chinese; The CEOs of Nvidia and AMD, the global AI chip industry leaders, are Chinese, and there are even more Chinese among well-known scientists in the field of AI in the United States.

According to Yang, China's advantage lies in its leading position in applying AI and high-speed broadband to improve business productivity. As long as there is a breakthrough in technology, China will evolve faster than the United States in terms of AI applications. At present, the OpenAI team is trying to connect 10 million GPUs together, while domestic Internet giants only have a few thousand or tens of thousands of GPUs in their hands. "China should coordinate Internet companies at the national level to concentrate limited GPUs for the research and development of AI products, and seize the time to let leading companies take the lead in realizing the mass production of China's independent high-performance AI GPUs." Yang Fei said.

In response to the current "big chaos" situation in the domestic large-scale model track, Kai-Fu Lee called on entrepreneurs to pay attention to TC-PMF (Technology-Cost Product-Market-Fit×), reject the ofo-style money-burning method, and let the large model use a healthy and benign ROI (return on investment) to store energy for a long-distance run, and rush to China's AI 2.0 reform. Kai-Fu Lee believes that AI inclusion in the Chinese market will inevitably come.

The battle for AI hegemony among tech giants has begun, and the new round of competition will only become more fierce and comprehensive, and the future is still full of suspense.

(Editor: Zhang Jingchao Review: Li Zhenghao Proofreader: Yan Jingning)