The first in China, the large-scale model generates 20 million yuan in half a year!

 Datawhale dry goods 

Direction: AI Large Model Business, Source: Qubit

In half a year, use the large model to generate 20 million yuan for the company !

This number comes from the 2023 semi-annual report released by 360, which is also the first financial report handed over after the company bet on the big model.

66507c0397847fb830092b85556d673d.png

The revenue data announced together in this interim performance report is also more eye-catching:

  • Total revenue 4.503 billion yuan

  • Revenue in the second quarter was 2.537 billion yuan, a year-on-year increase of 10.42% and a quarter-on-quarter increase of 29.08%

  • The security business achieved revenue of 908 million yuan, of which the revenue in the second quarter was 654 million yuan

Among the established revenue-generating businesses such as Internet advertising, security business, games, and smart hardware, the "360 Smart Brain" large model has not achieved the most revenue-generating results, but it has attracted special attention.

The reason for this is nothing more than the arrival of the era of large models, which has driven the entire AI application to "fly into the homes of ordinary people", behind which is the active layout and R&D exploration of manufacturers and startups.

Half a year later, the large-scale model track, which was full of life and death, finally put the results in black and white on the financial reports of major manufacturers.

360’s related business income of nearly 20 million yuan in the first half of the year seems to show that the company’s intensive AI strategy, with both wings flying together and scene layout, has worked.

The first domestic manufacturer that relies on large models for revenue

Let's first look at the specific figures announced by 360's financial report.

360's financial report for the first half of the year shows that during the reporting period, the company's total revenue was 4.503 billion yuan; the net profit loss attributable to listed shareholders was 231 million yuan, and the net loss in the same period last year was 398 million yuan, and the loss range has narrowed.

Revenue in the second quarter of this year was 2.537 billion yuan, a year-on-year increase of 10.42% and a month-on-month increase of 29.08%.

The growth of profit and revenue comes from the support of technology and related research and development investment.

The financial report shows that in the first half of this year, 360 company’s research and development expenses were 1.56 billion yuan, and the research and development expenses accounted for 34.64% of the revenue, an increase of nearly 1.21 percentage points and 3.59 percentage points compared with the first half of 2022 and the first half of 2021, respectively.

Heavy capital and high-intensity research and development have brought a certain amount of revenue feedback throughout the first half of the year.

This leads to the most noteworthy point in the 360 ​​financial report data, about its emerging business: the "360 Smart Brain" model has begun to generate income.

Amount of income: nearly 20 million yuan.

And "360 Smart Brain" officially opened the internal test time in April this year .

Nowadays, the global large-scale model industry is booming. In addition to basic user attention and user traffic, new products and services based on large-scale models may create completely new business models.

Relying on such a profit model, 360 may create even more amazing economic value by relying on large models in the future.

The "360 Smart Brain" large model is self-developed by 360, and it is positioned as a cognitive general artificial intelligence large model. Its capabilities include generation and creation, reading comprehension, and multiple rounds of dialogue. It is the first "trusted AIGC" approved by the Ministry of Industry and Information Technology in China Large Model Evaluation" large model.

Of course, a more important piece of information was released with the release of the semi-annual reports of several manufacturers:

360 is a large-scale model manufacturer in China, and it was the first to announce its revenue from large-scale models.

Relevant plans were disclosed in February, publicly displayed at the end of March, and the first search scene was launched in mid-April. As of June 30, the revenue of 20 million yuan was achieved. No matter the ability of the large model or the speed of 360, it is irrefutable. .

So here comes the question——

How did it achieve this result?

Break the curse of "burning money and not producing" large-scale models, why?

Promote multiple large-scale model businesses, and open APIs to seize various scenarios.

Observation found that 360's current businesses driven by large models are as follows:

  • Industry model : jointly launch "tax model" and "enterprise service model" with partners;

  • Urban large-scale model : Create city-level proprietary large-scale models for Sichuan, Chongqing and other cities, helping Sichuan and Chongqing to realize digital transformation and intelligent upgrade;

  • GPT Industry Alliance : In June, 360 launched an enterprise-level AI large-scale model solution and launched the GPT Industry Alliance;

  • Security Large Model : Released on August 9th, combined with 360’s past 15 years of AI security applications and security big data training, it is the first deliverable security industry large model in China.

So far, it has provided large-scale enterprise-level solutions for nearly 20 industries through API open model capabilities.

Constantly iterate the core large-scale model technology, launch available products, and implement various specific scenarios in thousands of industries at the same time, improve the efficiency and business boundaries of multiple industries, and help various industries realize value and enhance competitiveness.

This is probably the business experience that 360 can make a steady profit from large models.

And it is conceivable that the tentacles of 360 large models in the future will not stop at these existing directions.

c62bf7cb2c263b132dd4fa2c7eb3327a.png

Dry goods learning, like three times

Guess you like

Origin blog.csdn.net/Datawhale/article/details/132644602