The L’Oréal Groupe has announced a partnership with NVIDIA to advance the adoption of artificial intelligence (AI) across its beauty business. This collaboration is among L’Oréal’s recent investments in beauty technology, enhancing the company’s position at the forefront of digital beauty innovation. In this partnership, NVIDIA’s AI Enterprise platform will be used to accelerate the deployment of L’Oréal’s AI-powered beauty technologies at a wider and more efficient scale.
The rise of AI-powered beauty tech
McKinsey & Company projects the global beauty market to grow 5% annually through 2030, with online sales making up nearly a third of the total. In this fast-evolving landscape, beauty technology, powered by AI, AR/VR, and machine learning, has become a key differentiator, enabling brands to optimize operations and deliver more personalized, engaging customer experiences.
L’Oréal has been developing its beauty-related technologies for over a decade, including virtual try-on systems, personalized skincare tools, and agentic AI technologies. While these efforts have mostly centered on improving products and user experiences, the expansion of these systems has created a need for better digital infrastructure. L’Oréal’s partnership with NVIDIA is designed to meet that demand, paving the way for wider, more efficient deployment of its innovations.
NVIDIA’s AI Enterprise is a platform that enables companies to build and use AI tools at a faster and simpler rate. It handles the full AI process, from preparing data to training, testing, and deploying models, in one system. As L’Oréal integrates AI into tools like virtual try-ons and 3D renders, delivering them globally at scale becomes a key challenge. NVIDIA’s AI Enterprise offers the flexibility and power to support this rollout across cloud, data centers, and edge devices, while also opening new doors for NVIDIA in the consumer beauty market.
Also read: NVIDIA unveils new personal AI supercomputers
Inside L’Oréal’s technology patent portfolio
L’Oréal’s recent technology-related patent filings reflect the company’s strategy in developing its digital beauty systems, accompanying their transformation into a global beauty tech powerhouse.
L’Oréal: Patenting Activity
L’Oréal’s patent filings in audio-visual and computer technology saw their greatest increase between 2018 and 2019. This aligns with the company’s strategic shift in 2018, when it publicly declared its commitment to be at the forefront of the beauty tech industry. This shift came as digital campaigns became essential to beauty marketing. A key investment made by L’Oréal during this period was their acquisition of ModiFace, a company specializing in augmented reality and AI technologies for the beauty industry. This move laid the technical groundwork for some of L’Oréal’s future innovations in beauty tech.
L’Oréal’s surge in patent filings in 2020 reflects its growing focus on personalization, as highlighted in its annual report. A key example is Perso, an AI-powered device that creates custom cosmetic formulas, using ModiFace’s facial analysis technology.
The COVID-19 pandemic also accelerated demand for digital beauty solutions, leading to greater investment in virtual try-ons and AI tools. That same year, L’Oréal launched the Lancôme Shade Finder, which uses image analysis and AI to match foundation shades to individual skin tones.
L’Oréal: Top Legal Representatives
L’Oréal’s top legal representatives reflect a patenting strategy aligned with its key markets and R&D hubs. As a company that is headquartered in Clichy, Hauts-de-Seine, France, a couple of L’Oréal’s top legal representatives are based in Europe. Lavoix, a European IP firm based in France, handled the largest volume of L’Oréal’s filings in this domain. Aside from Lavoix, Casalonga, another French firm, also ranks among L’Oréal’s top representatives. Julian Mark Potter, a UK-based attorney from WP Thompson, has also handled a significant number of L’Oréal’s patents for audio-visual and computer technology.
In the United States, a substantial portion of filings were handled by Oblon, McClelland, Maier & Neustad and Christensen O’Connor Johnson Kindness. In China, filings are distributed across several firms such as China Patent Agent (H.K.), Patentsino, Patent Firm Shinmei Century, and Beijing Uni-Intel Patent & Trademark Office. In Japan, patent attorney Murayama Yasuhiko from Shiga International Patent Office represented the company in eleven filings, further supporting its presence in a key Asia-Pacific jurisdiction.
L’Oréal: Top Jurisdictions
In 2020, L’Oréal’s launched its Tech Accelerator program to drive innovation in data science, UX design, and technological engineering. With hubs in Paris, Shanghai, and most recently in New York, this helps explain the considerable number of audio-visual and computer technology patents distributed across France, China, and the United States.
Patent filings also reflect market size. The United States, recognized as the world’s largest beauty market, is L’Oréal’s top jurisdiction for patent filings in audio-visual and computer technologies as of 2025. China and Japan lead in the Asia-Pacific region, both in terms of their beauty market sizes and number of filed patents in this domain. By 2024, digital sales made up over 40% of China’s beauty market, 10% in Japan, and 50% in Korea. This highlights the growing need for digital innovation. L’Oréal China has also invested in employee training on tech and AI to support this shift.
L’Oréal: Top Technology Areas
The dominant areas in L’Oréal’s digital tech patent portfolio focus on media recognition, digital processing, and interactive consumer tools. Areas such as image and video recognition (G06V), processing (G06T), data recognition and presentation (G06K), and pictorial communication (H04N) highlight their capabilities related to their use of a customer’s visual information to determine personalized product recommendations. These technologies are central to ModiFace’s virtual makeup try-ons, the digital Lancôme Shade Finder, and the personalized product dispenser Perso.
Beyond enhancements and determining visual features, L’Oréal has also expanded its innovation pipeline to include data processing systems (G06F), specifically for commercial and marketing systems (G06Q), as well as computing arrangements based on computational models (G06N). These technologies support personalized marketing, inventory optimization, and content generation. Examples include CreaAItech, which uses generative AI for marketing visuals, and Demand Sensing, a forecasting tool for supply chains. This shows L’Oréal’s broader goal of applying tech across product development, operations, and consumer engagement.
It can be observed here that L’Oréal’s technologies mainly target the development of prototypes that will enable personalization and adaptive consumer experiences. The need to implement these on a larger scale through an AI management infrastructure is supplemented by their NVIDIA partnership.
L’Oréal’s Featured Patents
L’Oréal’s patents in audio-visual and computer technologies highlight their agenda towards delivering personalized and convenient beauty shopping experiences.
Identifying skin color in images with different lighting conditions
Online shopping offers convenience for customers as it provides them with a large catalog of product recommendations with just a click. However, customers may find in-person shopping to be more reliable for purchasing tone-dependent beauty products such as foundations. Although there are platforms that allow a person to upload an image to generate product recommendations, the image may not be an entirely reliable determinant of skin tone due to technical limitations, specifically lighting conditions.
U.S. Patent No. 11,191,342 presents a system for estimating the accurate skin tone from images or videos taken in environments with varied lighting conditions. An image or a video may be used as an input for this system. The input will then be transmitted to a skin color determination device, which uses one or more machine learning (ML) models to estimate the user’s skin color.
There are two possible setups that may be employed. The first setup involves two ML models: one that will estimate the lighting condition based on the image, and another that will use both the image and estimated lighting information from the first model to determine the user’s true skin color. This setup enables a more accurate detection of the actual color as it also considers lighting as a variable. Meanwhile, the second setup involves one ML model which will take the image and directly output the user’s estimated skin color, which reduces system complexity. The resulting skin color information from either setup will then be used to generate relevant product recommendations.
The patent, titled “Techniques for identifying skin color in images having uncontrolled lighting conditions”, was filed on July 18, 2019 and granted on December 7, 2021. The inventors are Christine Elfakhri, Florent Valceschini, Loic Tran, Matthieu Perrot, Robin Kips, and Emmanuel Malherbe.
Digital makeup palette for virtual try-on applications
Traditional virtual makeup try-on applications often lack options for customization and are limited in terms of their capabilities in adjusting to different facial features. They also often struggle to address problem areas, such as blemishes or hyperpigmentation, or emphasize facial features that the user may want to enhance through makeup. U.S. Patent No. 12,136,173 addresses these gaps through a comprehensive augmented reality (AR) makeup system.
The makeup system is composed of four different units. The makeup objective unit generates a list of predefined makeup looks based on different categories, such as “evening glam” or “spring”. This unit will also ask a user to indicate their experience level in terms of using makeup. Once a user has selected a makeup look, the makeup objective visualization unit will analyze a user’s facial features such as shape, skin tone, and landmarks. This unit will also take the lighting conditions of a user’s environment into account. Then, the makeup palette unit will provide a custom palette of products that aligns with a user’s intended makeup look, physical features, and lighting, such as shown in Fig. 6. The makeup objective visualization will then generate digital makeup filters corresponding to the selected products and overlay them in real time. Notably, the filters are customized to user data in order to account for differences in facial features and user-declared problem areas or emphasized zones.
Aside from the automatic try-on option via filters, the system also has a virtual try-on unit that allows users to try on applying the makeup through gestures. This unit interprets the gestures made by users, judges if it was applied correctly, and offers a “redo” if it is needed. This is beneficial for novice users who want to learn application techniques. The final look may be saved, shared, or even exported for live video platforms.
The patent, titled “Digital makeup palette”, was filed on December 30, 2020 and granted on November 5, 2024. The inventors are Mindy Christine Troutman, Sandrine Gadol, and Francesca D. Cruz.
Personalized product recommendations based on an image
Many customers get “look inspirations” from photos they see online or on social media. However, these images do not usually identify the details of the exact makeup product used. On the customer’s side, it becomes difficult to replicate the look. On the brand’s side, it becomes a missed opportunity to recommend similar looks and market their products.
U.S. Patent No. 11,521,013 uses a system of ML models to address the mentioned issues. First, the system receives a facial image from a user. The image will be processed using ML models that are trained to detect and identify cosmetic product information on specific facial regions. For example, one model may focus on the lips and identify that a “red glossy lipstick by Brand X” has been used. The combination of the products from the provided image will be grouped together as a single “look” entry, which will be stored in a database.
The system will then perform cluster analysis to group the new look entry with other looks based on the similarity of their products, textures, palettes, or trends. Based on the clusters, the system will then present the detected products and recommended alternatives back to the user. Other products that can complement the detected ones (e.g. a certain shade of blush to complement the detected lipstick product) may also be offered, thus simulating the in-person shopping experience of being recommended certain products by a salesperson based on your initial interest.
The patent, titled “Systems and methods for providing personalized product recommendations using deep learning”, was filed on May 11, 2021 and granted on December 6, 2022. The inventors are Grégoire Charraud, Helga Malaprade, Géraldine Thiebaut, Matthieu Perrot, and Robin Kips.
All featured patents were represented by Christensen O’Connor Johnson Kindness (COJK)/L’Oreal USA, with attorneys Brandon Stallman, Laura Cruz, and Runzhi Zhao et al. listed on the application.
Future outlook for beauty tech
Generative AI is projected to add $9 billion to $10 billion to the global economy based on its impact on the beauty industry. In their analysis, McKinsey & Company emphasizes that the gap between the laggards and leaders in the industry will further widen once leading groups successfully deploy genAI at scale. As beauty experiences become increasingly digital, the ability to deploy AI tools reliably across global markets becomes a competitive advantage. Aside from front-end applications, these tools may also be used to bolster other facets of the beauty business within their internal value chain.
Ultimately, this partnership reflects a strategic alignment wherein L’Oreal benefits from enterprise-grade AI deployment and NVIDIA gains the opportunity to demonstrate the versatility and reliability of its AI Enterprise platform for a consumer-facing industry. This collaboration sets a precedent for future cross-industry partnerships in which AI is not just a technological upgrade, but a critical component in demonstrating a company’s ability to adapt amidst rapidly-changing technologies and shifting consumer preferences.
