Core conclusion: We expect that in 2024-2026, Shangtang's revenue will be 4.5, 5.9, and 7.8 billion yuan, respectively, and net profit to mother will be -4, -2.8, and -1.5 billion yuan, respectively. We believe that Shangtang is a leading domestic artificial intelligence company. The generative AI business and smart car business have great potential for growth. It was covered for the first time and gave the company a “buy” rating.

Shangtang Technology is a leading global artificial intelligence company and a pioneer in commercializing generative AI. 1) From AI1.0 to AI 2.0, Shangtang continued to lead. The company was founded in 2014 and is a leader in the field of computer vision; in 2023, based on perception and decision intelligence in the AI 1.0 era, the company successfully made the leap to generative AI in the AI 2.0 era. 2) 2024H1, the company's losses narrowed further, and generative AI revenue continued to double, becoming the company's largest business. At the same time, the company has maintained a high investment in R&D in recent years, accounting for more than 70% of R&D personnel at the end of the first half of 2024.

The “big device+big model” forward-looking layout pioneered the closed loop of generative AI business. The company has deep accumulation at the hardware and software levels, and has a remarkable first-mover advantage. 1) Large installation: Shangtang has proactively deployed SenseCore. As of July 2024, the total operating power scale reached 20,000 PetaFlops, with over 0.054 million GPUs. According to IDC data, Shangtang currently ranks first in the domestic GenAI IaaS field. 2) Nisshin SenseNova large model: deep collaboration with large devices. The market-based model has excellent performance, and the interactive experience has been upgraded; the product matrix related to the big model is rich, and a “cloud-side” full-stack layout has been carried out to continuously explore AI implementation scenarios. According to IDC data, in 2023, Shangtang ranked second in China's largest model platform and application market share.

Shang Tang Zhuying: Building a “drive-cabin-cloud” three-in-one full-stack AGI technology architecture is expected to rank first in the domestic smart driving tier 1. 1) Intelligent driving: It has strong technical strength and innovation capabilities. UniAd, an end-to-end autonomous driving solution for mass production, has been deployed in actual vehicles, and the first smart driving model DriveAgi developed based on multi-modal large models has been launched. 2) Smart cockpit: Providing one-stop cockpit solutions with DMS and OMS as the leading products, the cockpit AI software market share ranked first for 5 consecutive years.

3) AI cloud: based on large devices to maximize data value.

Risk warning: AI technology breakthroughs fall short of expectations, large-scale model commercialization process falls short of expectations, increased industry competition, and risk of bad accounts receivable.