How Uniqlo and Fast Retailing use AI and data in the Ariake Project

KEY TAKEAWAYS Summary by the editors

1. Fast Retailing, the parent of Uniqlo, launched the Ariake Project in 2017 to become what it calls a digital consumer retailing company that produces, transports and sells the exact volume of products customers want, when they want them.
2. In September 2018 Fast Retailing announced a partnership with Google Cloud covering demand forecasting, data access and collaboration tools for the Ariake Project.
3. Fast Retailing's automated Ariake distribution centre in Tokyo, built with Daifuku, began full operation in October 2018 with a reported 90 percent reduction in staff.
4. In its Integrated Report 2025 Fast Retailing says it uses AI to significantly increase the accuracy and speed of customer feedback analysis and dissemination.
5. Fast Retailing has published few quantified results for AI specifically, so the effect of AI on forecast accuracy or inventory levels cannot be measured from public sources.

Uniqlo's parent Fast Retailing uses AI and data mainly to run a demand-driven supply chain under its Ariake Project, launched in 2017. The programme connects customer feedback, sales and inventory data with planning, factories and automated warehouses, with the stated goal to "produce, transport, and sell the exact volume of the exact products that customers want, exactly when they want them". AI is one tool within this information system rather than the centre of it.

What has Fast Retailing built?

The Ariake Project is named after the Tokyo district where Fast Retailing opened an office and a highly automated warehouse. According to the company's Integrated Report 2025, it was launched in 2017 to reform operational management and create "a new type of digital consumer retailing company". The building blocks it describes are:

• Shared information: since 2023, worldwide systems that visualise sales, inventory levels and customer feedback by individual store in real time.
• Integrated planning: linking production, sales and distribution plans, sharing information with factories, allocating warehouse stock and transporting goods to stores.
• AI for customer feedback: faster and more accurate analysis and distribution of customer comments to the teams that act on them.
• Automated warehouses: starting with Ariake in Tokyo and including a 110,000 square metre automated warehouse in the Netherlands where products are automatically picked, packed and sorted.

Why did Fast Retailing invest in AI?

Uniqlo sells a relatively narrow range of basics in very large volumes, so forecasting errors are expensive in both directions: excess stock ties up cash and leads to markdowns, while shortages lose sales of core items. When Fast Retailing announced its partnership with Google Cloud in 2018, the goals included translating customer feedback into products faster, moving away from a "make it and they will come" approach, improving communication with factories and building more efficient distribution.

CEO Tadashi Yanai framed the project as an organisational change as much as a technical one: "Making information accessible to all our employees is one of the foundations of the Ariake project, because it empowers them to use human traits like logic, judgment, and empathy to make decisions."

Read also

How Inditex uses AI and data: RFID and a demand-driven supply chain

How does it work (data, models, process)?

Fast Retailing has not published model details, but its descriptions show how data flows through the chain:

The Google Cloud partnership gave Fast Retailing access to Google's Advanced Solutions Lab in Tokyo, collaboration tools and support for demand forecasting and data accessibility. Yanai said in 2018 that the work had gone "well beyond demand forecasting" and changed how teams work together. In logistics, the Ariake warehouse built with Daifuku transfers products from trucks, reads electronic tags and checks inventory with little human involvement, allowing round-the-clock operation.

Analysis: the report also cites a 1993 memo stating "Business = Systems", stressing that operations must be standardised before they are digitalised. That sequence is the less visible part of the story: AI forecasting only helps when product, store and inventory data are consistent across the organisation.

What results has Fast Retailing reported?

The most specific published result is in logistics. Fast Retailing reported a 90 percent reduction in staff at the Ariake distribution centre after automation, and in October 2018 Retail TouchPoints reported a planned investment of 885 million dollars in warehouse automation with Daifuku. Yanai said at the time: "We want to introduce automated warehouses around the world as soon as possible."

For AI in planning and design, the company describes qualitative improvements, such as more accurate and faster analysis of customer feedback, but it has not disclosed forecast accuracy, inventory reductions or margin effects attributable to AI. Fast Retailing's financial performance reflects many factors, from product strategy to store expansion, and public sources do not allow the contribution of the Ariake Project to be isolated.

What are the limits and open questions?

The warehouse automation figure invites the obvious question of what happened to the jobs, and of how the model translates to markets with different labour costs. Large automation programmes also require heavy upfront capital and take years to pay back. On the data side, demand forecasting for basics benefits from long, stable sales histories; it is less clear how well the same methods work for fashion-led or newly launched items. Finally, the term "Ariake Project" covers a broad transformation, which makes it difficult for outsiders to distinguish AI from process redesign and organisational change.

Read also

How Levi Strauss & Co. uses AI: the data programme and its lessons

What can other fashion companies learn?

Analysis: few companies can match Fast Retailing's scale or investment budget, but the principles behind the Ariake Project apply to brands of any size, because most of them concern data discipline and organisation rather than advanced algorithms:

1. Standardise before you automate. Consistent product, store and inventory data is a precondition for useful forecasting.
2. Make data visible to people. Fast Retailing presents information access for employees as the foundation, with AI supporting decisions rather than replacing them.
3. Close the loop with customer feedback. Systematically analysing customer comments and routing them to product teams is a practical, lower-risk AI use case.
4. Treat logistics as part of the AI case. The clearest measurable gains so far are in automated warehouses, not in algorithms alone.

Frequently asked questions

What is Uniqlo's Ariake Project?

The Ariake Project is Fast Retailing's transformation programme, launched in 2017, to become a digital consumer retailer. Its aim is to produce, transport and sell exactly the products customers want, when they want them, by connecting customer, sales, inventory, factory and logistics data.

Does Uniqlo use AI for demand forecasting?

Yes. Fast Retailing partnered with Google Cloud in 2018 on demand forecasting and data access as part of the Ariake Project. The company has not published forecast accuracy figures.

How does Fast Retailing use AI with customer feedback?

According to its Integrated Report 2025, Fast Retailing uses AI to significantly increase the accuracy and speed of analysing and sharing customer feedback, so product and store teams can respond in a targeted way.

How automated are Uniqlo's warehouses?

Fast Retailing's Ariake warehouse in Tokyo, developed with Daifuku, began full operation in October 2018 and was reported to need 90 percent fewer staff. The group also operates a 110,000 square metre automated warehouse in the Netherlands.

GuideThe complete guide to AI in the fashion supply chain and sustainabilityRead the complete guide