OpenAI expands Singapore hub as it searches Asia for data centers to power ChatGPT

Singapore anchors a regional expansion

OpenAI plans to grow its Singapore team to roughly 50 to 70 people by the end of 2025 as it searches Asia Pacific for places to build the data centers behind ChatGPT. The expansion, from a small group when the office opened, puts Singapore at the center of a regional strategy combining business partnerships, developer support and talks with governments.

As of October 15, 2025, ChatGPT has about 700 million weekly active users. Serving that audience requires vast amounts of computing capacity, specialized chips and reliable electricity. OpenAI is pursuing both a larger presence near its Asian customers and more infrastructure to support its products.

Brad Lightcap, OpenAI's chief operating officer, has said the company will keep investing in line with demand and is open to further expansion in Singapore or elsewhere as consumer use and enterprise adoption grow. Asia Pacific is among its fastest growing regions.

Singapore stands out for its use of ChatGPT relative to its population. An estimated one in four residents uses the service in a country of 5.9 million people, placing it among OpenAI's leading markets for adoption per person. That level of engagement gives the company a strong reason to deepen its local presence.

The immediate commitment concerns staff and regional operations. OpenAI is also exploring data center locations across Asia, with Singapore among the places being considered for regional capacity. No confirmed Singapore data center site or opening date has been given. Its role as a business hub is developing while decisions about where to put additional computing equipment remain open.

What the larger Singapore team will do

OpenAI expects a bigger local team to help it work more closely with governments, enterprises and developers across the region. Its executives say that presence will support products suited to local languages and industries, stronger developer relations and pilot projects with public agencies.

Singapore Airlines is already among its prominent customers. The airline is integrating advanced prompts and assistants into work involving staff and customer interactions. Regional technology companies, including businesses in ride hailing and electronic commerce, have also been experimenting with OpenAI models to personalize services, produce content more quickly and improve customer support.

OpenAI's Singapore expansion is expected to bring roles in partnerships, developer relations, policy, research and customer engineering. Those functions can help local companies move from small trials to systems used in everyday operations. They also create opportunities for engineers based in Singapore to help develop features and reference solutions that could later reach users elsewhere.

Singapore offers access to multinational headquarters and a concentrated community of regional developers. OpenAI's leaders cite both as reasons for establishing a local hub. More than four million developers globally have built applications using its technology, creating a large network of people who can provide feedback on how the models work in practice.

A local team can shorten the time between a customer identifying a problem and the company understanding it. For Asian businesses, that could mean closer attention to regional languages, industry needs and the way employees carry out their work.

Language resources and public partnerships

OpenAI has partnered with AI Singapore, the national research and innovation program, to support open datasets for Southeast Asian languages and improve the region's representation in large language models. These models are the systems behind tools that generate and interpret language.

AI Singapore has described the partnership as a way to strengthen regional language resources and give local teams better access to advanced tools. Open datasets provide material that teams can use when developing and assessing language technology, making them part of the effort to serve multilingual societies.

Singapore officials have pointed to the country's investments in computing resources, its talent base and programs promoting responsible AI. They see the arrival of a company that conducts research and builds its own frontier models as support for the country's technology ecosystem.

Leaders from Singapore's economic and digital agencies have said OpenAI's presence can create opportunities for local startups and deepen cooperation on computing capacity, safety and skills. Public agencies have also engaged on practical matters including safeguards, data governance and workforce training.

OpenAI has held discussions with agencies responsible for industry development, research funding and digital trust. Officials have spoken about connecting access to computing resources with safety evaluations and skills training, so that adoption includes both technical capability and responsible use.

Why ChatGPT needs so much infrastructure

ChatGPT's roughly 700 million weekly users help explain OpenAI's search for more computing capacity. Each prompt triggers complex calculations across chips. Repeating those calculations for a vast audience requires much more than the software people see on their screens.

Training large models and running them for users require tens of thousands of advanced graphics processors, large amounts of memory, fast networks and stable power. Major installations can need electricity measured in hundreds of megawatts or even gigawatts. As demand grows, computing capacity affects how quickly OpenAI can introduce products and features.

Sam Altman, OpenAI's chief executive, described that connection at a recent press event focused on infrastructure. He said the company intended to move quickly because access to computing resources shapes what it can deliver.

We will push on infrastructure as hard as possible because it will drive our ability to deliver technology and products.

OpenAI has outlined an infrastructure expansion extending over several years. In the United States, it is partnering with Oracle and SoftBank on sites in Texas, New Mexico and Ohio, with an eventual capacity target near 7 gigawatts. The projects are among the largest announced during the current AI expansion.

Clay Magouyrk, a senior Oracle executive who leads Oracle Cloud Infrastructure, has said campuses costing billions of dollars require technology providers and investors to work together. He described the coordination needed across the systems that make a large AI facility possible.

This work is not possible for any one company to do alone. It takes close coordination across compute, networks, power and finance.

Oracle's involvement illustrates the range of partners behind the buildout. Servers and chips must be supported by networks, electricity supplies, construction and financing. OpenAI's expansion depends on those parts becoming available together.

Asia joins the search for new sites

Jason Kwon, OpenAI's chief strategy officer, has been meeting leaders across Asia Pacific to explore data center sites and partnerships. His itinerary includes Japan, South Korea, Australia, India and Singapore.

The discussions form part of OpenAI for Countries, an initiative inviting governments to invest jointly in sovereign AI capacity, develop local talent and customize services for national languages and needs. Sovereign AI capacity, in this context, means computing resources and services developed around a country's own requirements, including expectations about where data is processed.

OpenAI has indicated interest in as many as 10 new large data center projects worldwide through the initiative. More than 30 countries have expressed interest. Discussions in Asia have focused on bringing computing resources closer to users and meeting the data sovereignty expectations of regulators and enterprises.

The first international project is in the United Arab Emirates. OpenAI and its partners are building a large cluster in Abu Dhabi under the Stargate initiative, backed by local developer G42 and global technology partners. The project starts with a phase of 1 gigawatt and has plans to grow several times larger.

India is one of OpenAI's largest markets by users. Industry discussion has pointed to the possibility of a site with 1 gigawatt of capacity there. During a visit, Kwon connected the need for infrastructure with the need to develop people's skills.

In the brief time that I have spent in India, it is clear that the country's leadership understands that maximizing AI's benefits requires significant investments in two areas: Core infrastructure and cultivating AI talent. By leading in these areas and empowering people to harness frontier intelligence, India can accelerate economic growth.

Japan and South Korea have established chip and data center ecosystems. Australia offers land and renewable power resources that could support large installations connected to the electricity grid. Singapore brings an established data center cluster and a policy focus on growth that accounts for environmental demands.

The expansion spans several stages, with the end of 2025 providing the stated deadline for Singapore's staffing plans:

• Earlier in the decade, Singapore paused new data center permits before introducing a framework favoring more efficient facilities.
• By October 15, 2025, OpenAI is expanding its Singapore presence and exploring infrastructure partnerships across Asia Pacific.
• By the end of 2025, OpenAI plans to have roughly 50 to 70 staff in Singapore.
• Over several years, OpenAI and its partners plan additional capacity in the United States and Abu Dhabi while assessing further international projects.

What changes for Singapore businesses

Singapore's attraction extends beyond consumer use of ChatGPT. Its dense digital infrastructure, strong international connections and mature data center sector support companies operating across national borders. The government has also set out frameworks for testing and deploying AI.

OpenAI already works with businesses in transport, tourism and gaming, alongside Singapore Airlines. A larger local presence gives those companies more direct access to people who can help adapt models to their services and operations.

Telecommunications companies are working with OpenAI to put conversational assistants into customer channels and network operations. Travel and hospitality groups are testing AI for itinerary planning, service recovery and marketing content. Banks and insurers are developing copilots to help staff summarize documents and flag risks.

For employees, these projects bring AI into specific tasks rather than leaving it as a separate chatbot. An assistant might support customer service, while a copilot helps a worker review documents. Successful deployments also expand the pool of people with practical experience in designing prompts, fine tuning models and checking outputs responsibly.

Singapore's industry partners expect benefits to extend beyond a single company's office. More local collaboration can connect startups, established businesses, researchers and public agencies around access to computing resources, skills and safety testing.

How Microsoft and other partners fit

Microsoft remains central to the way OpenAI's models reach enterprises worldwide. The longstanding partnership combines Microsoft's cloud infrastructure, security and enterprise sales network through Azure with OpenAI's model research and product development.

OpenAI is also engaging customers directly in Singapore and across Asia. Some clients want closer contact with its researchers or access to advanced features beyond standard packages. Its regional expansion supports those relationships alongside the Microsoft partnership.

Large deployments can involve several providers. Oracle supplies cloud capacity to Stargate projects, while chipmakers such as Nvidia are deeply involved in the supply chain. A customer using an AI service may therefore depend on a wider network of infrastructure and technology companies.

For enterprise buyers, the practical concerns are reliability, cost, latency and compliance with data rules. Latency is the delay before a system responds. These priorities help explain why OpenAI is pursuing both direct customer relationships and sales through partners while seeking computing capacity closer to users.

Electricity and land shape Singapore's options

Singapore's established technology sector makes it attractive, but land and electricity are finite. OpenAI is considering regional capacity either inside the country or through nearby campuses connected by fast fiber networks.

Such an arrangement would let Singapore serve as a center for operations and partnerships while some computing equipment sits in neighboring markets. This model is already visible in projects in Malaysia and Indonesia that serve businesses and users based in Singapore through connections with little delay.

Singapore's approach to data centers has balanced expansion with sustainability. After the earlier pause on new permits, the government introduced a framework favoring energy efficient designs, district cooling and operations that can increasingly use power with lower carbon emissions.

Several operators are testing advanced liquid cooling and densely packed server racks. These approaches aim to put more computing capacity into less space while reducing wasted energy. Cooling is part of the infrastructure challenge alongside the electricity needed to run the chips themselves.

Large AI projects also need plans for long term electricity purchases, energy storage and participation in green import corridors intended to bring renewable power from neighboring countries in Southeast Asia. Water conservation, reuse of heat and recycling server components are further operating considerations.

Safety rules and constraints on construction

Singapore's regulators have encouraged transparent reporting so customers can compare the environmental performance of data centers. The country has also developed practical AI governance toolkits and is home to one of the world's first national AI safety institutes.

Companies deploying chatbots in sensitive areas are expected to test for bias, privacy problems and security vulnerabilities. OpenAI has been engaging on those issues while developing safeguards for its consumer and enterprise products.

Advanced chips remain a bottleneck for expansion. Power distribution equipment can take a long time to arrive, and construction of very large sites can encounter land use and permit constraints. These factors affect how quickly announced capacity becomes usable infrastructure.

Governments also have different requirements for sensitive data. Some want it processed within their borders, while others are considering rules governing the training and evaluation of frontier models. Those choices influence which locations and partnerships can serve particular customers.

OpenAI's Singapore strategy connects two needs: a larger team near a fast growing market and more computing capacity to support it. Singapore offers networks, talent, customer demand and established policy frameworks. The location and pace of regional data center construction will also depend on power, chips, land and regulatory decisions.

At a Glance

• OpenAI plans to expand its Singapore team to roughly 50 to 70 people by the end of 2025.
• ChatGPT has about 700 million weekly active users, with an estimated one in four Singapore residents using it.
• Singapore Airlines and regional businesses are using or testing OpenAI technology across customer service and staff operations.
• OpenAI is working with AI Singapore on open datasets for Southeast Asian languages.
• Jason Kwon is exploring infrastructure partnerships in Japan, South Korea, Australia, India and Singapore.
• United States projects with Oracle and SoftBank target capacity near 7 gigawatts; Abu Dhabi starts with 1 gigawatt and plans further growth.
• Microsoft remains a central enterprise partner while OpenAI also develops direct customer relationships.
• Singapore's regional role could include nearby data centers connected by fast fiber, alongside its expanding local office.
• Electricity, chip supply, permits, environmental requirements and data rules will shape the pace of construction.