美妆“金饭碗” 正在被AI蒸馏

This article explores how artificial intelligence has penetrated upstream R&D in the beauty industry, bringing disruptive changes to product development. Key takeaways are as follows:

1. Traditional beauty R&D relies on manual experimental screening, delivering results largely by chance, taking months or even years at high cost. By summarizing patterns in historical data, AI can conduct targeted searches for optimal solutions, cutting R&D cycles roughly in half and substantially reducing trial-and-error costs to date.

2. AI is already being applied in real-world operations: brands including Proya, Huanya Group and Zhimei Group have integrated AI-developed ingredients into their star products. Multiple AI-powered beauty R&D startups have also raised hundreds of millions of yuan in funding, reflecting strong investor confidence in this segment.

3. Ordinary R&D practitioners do not need to be overly anxious about job displacement. By treating AI as a tool and collaborator, learning to work with the technology and becoming "AI-native," they can adapt to the new working model.

AI’s entry into upstream beauty R&D has created unprecedented opportunities for beauty brands to accelerate product development and boost competitiveness. Key insights are as follows:

1. For product R&D, AI enables brands to complete virtual screening of ingredients in advance, transforming the traditional broad-net screening model of contract research organizations (CROs) into a targeted approach. This can cut hundreds of millions of yuan in R&D costs, halve development cycles, and allow brands to respond faster to market demand for ingredient upgrades.

2. There are already multiple proven successful use cases: Proya used AI to complete target screening for Cyclopeptide-163, which is now used in its popular Ruby series; Zhimei Group used AI to identify bakuchiol, a gentle alternative to retinol, and successfully built it into a top-selling product.

3. Today, core demands of leading beauty brands for raw materials are differentiation, exclusivity and safety, all of which AI can match quickly. International giants including L’Oréal and Shiseido have already quietly built out AI R&D capabilities, and Chinese domestic brands can also leverage this opportunity to break through innovation bottlenecks and achieve an overtake on the curve.

AI’s penetration into upstream beauty R&D brings new growth opportunities for beauty sellers, along with notable risks to watch for. Key takeaways are as follows:

1. On the opportunity side, consumers now demand faster ingredient iteration for beauty products, and increasingly favor products with clearly defined efficacy and innovative ingredients. AI enables sellers to quickly launch differentiated new products that meet market demand and capture greater market share.

2. Mature cooperation models are already available: sellers can partner with AI R&D firms, raw material factories and CROs to screen high-potential ingredients via AI first before conducting downstream validation, which greatly cuts trial-and-error costs. Multiple Chinese domestic brands have already posted successful results from this approach.

3. Key risks to note: AI implementation still faces clear bottlenecks. Back-end wet-lab validation still relies heavily on manual work, no industry player is willing to share core R&D data, and regulatory approval for entirely new raw materials takes more than one year. Sellers should avoid jumping on the bandwagon of concept hype, and map out a clear implementation path before committing resources.

AI has brought new competitive opportunities and a clear digital transformation direction for beauty raw material factories. Key insights are as follows:

1. For R&D and product design, traditional raw material development relies on chance to identify high-quality sequences, resulting in high uncertainty and slow iteration. AI can leverage historical data to identify patterns, enabling rational, targeted R&D that greatly shortens development cycles and cuts trial-and-error costs. For example, Meishangjie Biotech has used AI to break out from the shadow of industry giants and raised tens of millions of yuan in funding.

2. For business opportunities, differentiated raw materials developed via AI are highly sought after by leading brands. Some players have already transformed from offering only R&D services to a combined "R&D + in-house raw material supply" model, expanding both revenue streams and business boundaries.

3. Key takeaways for transformation: factories can proactively partner with AI service providers to build dedicated AI R&D platforms, accumulate positive and negative experimental data from their own work to continuously optimize models, and core teams need to proactively learn AI-related knowledge to adapt to the new R&D model.

AI-powered beauty R&D brings new development opportunities and strategic adjustment directions for all types of beauty-related service providers. Key insights are as follows:

1. Industry trend: AI is reshaping industrial division of labor in beauty R&D. It has expanded from downstream customer service and marketing to upstream R&D, and the traditional boundaries between AI service providers, raw material factories and CROs are blurring, with a new industrial structure gradually emerging.

2. Core customer pain points: For beauty brands, the biggest pain points are high costs and long cycles of traditional R&D, which make it hard to deliver the differentiated innovative ingredients the market demands. Brands need to cut costs and speed up development while ensuring the safety and compliance of raw materials.

3. Solutions and development directions: Traditional CROs can proactively integrate AI technology to upgrade their existing services. AI service providers can expand their business boundaries, moving from pure R&D service offerings to combined R&D and raw material supply. They can also explore partnerships with automated laboratories to resolve the pain point of delayed back-end wet-lab validation. They can also transform their revenue model from project-based fixed fees to sales-based royalties, boosting profit margins.

The wave of AI-powered beauty R&D brings new development opportunities for platforms, alongside risks that need to be managed. Key takeaways are as follows:

1. There is currently strong industry demand for AI-powered beauty R&D, with a growing number of beauty brands, raw material factories and AI R&D firms building out capabilities in this space, all requiring platforms to connect upstream and downstream resources. Platforms can build new service segments focused on AI beauty R&D to attract innovative market participants to join.

2. Latest industry developments show that capital is already highly focused on AI for Science beauty startups, with multiple players raising large funding rounds. Both leading international and Chinese domestic beauty brands have already launched AI R&D projects. Platforms can carry out targeted investment promotion to attract innovative AI R&D and AI-enabled raw material companies, enriching the platform’s product categories and business ecosystem.

3. Risks to avoid: AI beauty R&D still faces many implementation bottlenecks, including a mismatch between overhyped cloud-based solutions and limited real-world physical testing, lack of core R&D data sharing, long new raw material approval cycles, and widespread concept hype. Platforms should verify implementation progress when onboarding related companies to avoid risks from false marketing claims.

The penetration of AI into upstream beauty R&D is a new industry development, with many new problems and business models that merit further research. Key insights are as follows:

1. New industry trends: Previously, AI mainly impacted downstream links of the beauty industry such as customer service, design and marketing; it has now moved into the higher-barrier upstream R&D segment, restructuring the core logic of beauty development. The traditional broad-net, chance-based model is being replaced by precise, rational design. Multiple leading brands have already built out AI R&D capabilities, AI-focused companies have received substantial capital inflows, and the industry is poised for a major shakeout.

2. Unresolved industry challenges: AI implementation currently faces a mismatch between overheated cloud-based development and underdeveloped real-world physical testing, with back-end wet-lab validation remaining a major bottleneck. Industry players are unwilling to share core R&D data, small and medium-sized market participants lack sufficient data to train AI models, and new raw material approval takes more than one year—all of these factors constrain industry innovation.

3. Emerging business models: AI service providers have already explored a transformation path from pure R&D service offerings to a combined "R&D + in-house raw material supply" model, shifting from project-based fixed fees to sales-based royalties. The boundaries between traditional industry players are gradually blurring, and a new division of labor is taking shape—all of which require continued tracking and research.