Once the silent engine behind the world’s electronics, Foxconn is stepping into the spotlight, betting big on becoming a platform powerhouse for the intelligent systems age.

With more than US$200 billion in annual revenue and a global workforce topping 900,000 during peak season, Hon Hai Precision Industry Co. is a giant by any measure. Yet its origins are far more modest. The company was founded in 1974 with capital of just US$7,500 and fewer than a dozen employees by Terry Gou, now a billionaire businessman, philanthropist, and one-time political figure.

Better known globally as Foxconn, the Group began in what is now New Taipei City as a small operation making plastic parts for television sets. Its first major pivot came in the 1980s, when it shifted to producing computer connectors, an early sign of the adaptability that would come to define it.

The shift proved highly lucrative. Foxconn evolved into one of the world’s largest contract electronics manufacturers, assembling devices for global giants like Apple, Dell, and Sony. With hundreds of factories and a workforce in the hundreds of thousands, the company became a critical – if largely invisible – pillar of the digital age.

Today, Foxconn is undergoing another reinvention, this time as a platform service company aiming to build the infrastructure and applications that power artificial intelligence. It’s a bold bet that reflects not only its appetite for transformation, but also Taiwan’s broader ambition to remain indispensable in a rapidly evolving tech landscape.

That ambition was on full display at Nvidia’s GTC 2025 conference, held mid-March in San Jose, California. At the global AI conference, Foxconn showcased AI-driven initiatives in smart manufacturing, urban infrastructure, digital health, and open-source innovation. The presence of its chairman, Young Liu, who led a 75-person delegation of engineers and executives, reflected the strategic weight Foxconn places on its AI transition.

“We’re in a very important position to help the Group upgrade manufacturing,” says Jimmy Huang, Foxconn deputy spokesperson. He adds that cross-functional teams are now active across Foxconn’s facilities in the United States, Mexico, and China, identifying where AI can bring measurable gains in efficiency and automation. The effort is part of a larger rethinking of what manufacturing can become in the age of intelligent machines.

Robot colleagues

Central to Foxconn’s transformation is a new generation of “AI Factories,” where automation, robotics, and machine learning work alongside people to deliver greater efficiency and adaptability.

“We are transitioning from a manufacturing service company to a platform solution provider,” says Shi Zhe, Foxconn’s chief digital officer, who is leading the company’s digital upgrade. “We have deep expertise in automation and manufacturing. Now we’re building AI models and platforms to accelerate digital transformation and support our workforce.”

Rather than seeking full automation at the expense of jobs, Foxconn is embracing a collaborative model in which AI functions as a coworker. This approach, Shi says, can help address labor shortages and rising wage pressures in manufacturing hubs. Tools like Factory GPT, a proprietary suite of generative AI applications, support this effort by helping technicians troubleshoot problems, optimize workflows, and quickly adapt to new tasks on the production line.

Among the more futuristic displays at GTC was a humanoid robot developed by Foxconn Industrial Internet (Fii), a core subsidiary of the Group, designed to take on tasks that are too unpredictable for traditional robotic arms. Its presence in the Robotics Zone offered a glimpse into the company’s vision for next-generation manufacturing, where flexibility, mobility, and perception are just as important as speed and scale.

The long-term goal is for factories to mirror the adaptability of human workers while maintaining the uptime and precision of machines. Leo Guo, general manager of the Fii robotics business unit, sees humanoid robots as playing a critical role in this evolution.

“If we can introduce humanoid robots, then we can replace some workstations that currently cannot be automated,” he says.

Foxconn has already made significant headway in realizing its AI factory vision. Eight of its facilities have been recognized as Lighthouse Factories by the World Economic Forum – designated for their pioneering use of Fourth Industrial Revolution technologies. An additional 80 internal sites, known as corporate Lighthouse Factories, serve as experimental zones for piloting smart manufacturing systems. Across these facilities, Foxconn has deployed AI-powered tools for such tasks as automated optical inspection and predictive maintenance to reduce downtime and boost production quality.

Already the world’s largest producer of high-end servers, Foxconn plays a growing role in constructing large-scale AI computing clusters. It is one of a select few global manufacturers designated by Nvidia as a pilot-build-request, or PBR, partner for next-generation systems, thanks to its proven capabilities in hardware engineering and systems integration.

Such qualifications are increasingly vital in the AI era, where server reliability and efficiency can determine the speed at which models are trained and deployed. Supporting this momentum is Ingrasys, another Foxconn subsidiary, that develops liquid-cooling systems tailored for AI workloads.

On March 14, Foxconn Chairman Liu told media that he expects AI servers to be an NT$1 trillion (nearly US$32 billion) business this year, adding that the company’s goal is to continue increasing its market share to over 40%. Foxconn’s revenue from AI servers jumped 150% year-over-year (YoY) in 2024, while its overall server business grew 78%.

From simulation to realization

Digital twin technology also lies at the heart of Foxconn’s next-generation factory strategy. High-fidelity virtual replicas of production environments allow engineers to simulate, test, and refine complex systems before deploying them in the real world.

“Digital twin technology is very crucial and helpful for our global manufacturing deployment,” says Fii’s Guo. “We can model the entire production process, from robotic arm motions to full assembly lines, and optimize everything virtually before any equipment is installed.”

Using a growing library of pre-built components and AI toolkits, Foxconn engineers can redesign factory layouts in minutes and run simultaneous simulations to assess efficiency, ergonomics, and system resilience. This virtual-first approach has accelerated the rollout of highly automated or even fully “lights-out” production lines that operate with minimal human intervention, often in darkened facilities to conserve energy.

Yet bridging simulation and real-world execution remains a challenge. To succeed, the parameters of the simulation environment must be extremely detailed and highly accurate. “Without this level of precision, the simulation is not very useful,” says Guo. “A project like this really takes highly skilled engineering professionals and a significant investment of time.”

Foxconn has already begun offering its digital twin expertise to external clients, applying lessons learned from its Lighthouse Factories to help partners in sectors such as automotive and electronics implement AI-enabled smart manufacturing systems.

At the core of these advancements lies a new kind of infrastructure: intelligence itself. To power its smart platforms, Foxconn has developed its own large language model – FoxBrain.

Developed by the Hon Hai Research Institute, FoxBrain is a 70-billion-parameter model, incorporating advanced logical reasoning capabilities to better reflect how users in Taiwan communicate.

With this effort, Foxconn joins a select group of global tech firms – including Google and OpenAI – that are training proprietary AI models. The model was trained over four weeks on 120 Nvidia H100 GPUs via the Taipei-1 Supercomputer, using Nvidia’s NeMo framework.

Foxconn sees FoxBrain as more than a research milestone. The model is expected to become a core engine powering the company’s three strategic platforms: Smart Manufacturing, Smart Electric Vehicle, and Smart City. Internally, it could enable technicians to engage with complex systems in Traditional Chinese, analyze data, and streamline decision-making.

The project also marks a shift in Foxconn’s traditionally proprietary approach. The company is taking steps to engage more openly with developers, startups, and research partners, sharing elements of its work to broaden applications and promote collaboration.

Smart urban infrastructure

With foundational technologies like FoxBrain in place, Foxconn is extending its AI capabilities beyond the factory floor into the cities of tomorrow. At GTC 2025, the company showcased CityGPT, a generative AI platform designed to help modernize urban infrastructure and public services. Built on the same digital twin technology used in its factories, CityGPT creates virtual replicas of city environments to support tasks like traffic management, urban planning, and public safety.

Foxconn sees platforms like CityGPT as essential to the future of urban governance. At the conference, Christie Lin, who leads the company’s smart city initiatives, underscored the importance of “adaptive technology frameworks” that allow cities to rapidly prototype and implement AI-powered solutions.

One demonstration showed how real-time video feeds, combined with AI-driven analytics, can detect near-miss traffic incidents and recommend changes to road design or signaling. The system, powered by Nvidia’s Metropolis computer vision platform, is already being piloted to address pedestrian safety – a growing concern in cities across Taiwan, which in 2023 was described by international media as a “pedestrian hell.”

To support city officials, Foxconn is testing virtual AI assistants. One example, showcased at its GTC booth, was a digital traffic officer avatar designed to interface with urban planners and residents. This project aims to make data-heavy municipal decision-making more intuitive, timely, and responsive.

The smart city initiative is gaining traction. Foxconn recently signed an agreement with the state of Sonora, Mexico, extending for the first time outside of Taiwan the key intelligent platform after debuting it in Kaohsiung, while also joining the recent Taipei City’s Smart City Summit to explore real-world applications.

AI with a human touch

Among Foxconn’s most people-focused AI ventures is its entry into digital health. At GTC 2025, the company introduced Nurabot – a compact, wheeled service robot with expressive eyes and a gentle design intended to function as a hospital assistant rather than a clinical device.

Alice Lin, Director of UX/UI Design at Foxconn and a lead on the project, says the collaborative robot was developed to alleviate a growing strain on nursing staff. “Nurses often spend a significant amount of time walking between nursing stations and patient rooms to transport specimens and medications,” she notes. “While these tasks are essential, they’re time-consuming and physically demanding.”

Nurabot is designed to take on precisely those logistical duties. Using AI-powered navigation and onboard sensors, it can autonomously move through hospital corridors, avoid obstacles, and even operate elevators, freeing up nurses to focus on direct patient care. Its core appeal lies in its potential to ease a longstanding pain point in healthcare systems worldwide: overburdened staff stretched thin by routine responsibilities.

To prepare Nurabot for real-world deployment, Foxconn created Taiwan’s first-ever high-fidelity digital twin of an entire hospital ward, allowing its teams to simulate daily operations and fine-tune integration before introducing the robot into live settings. In an on-demand session at GTC, executives showcased a virtual replica of Nursing Station 106 at Taichung Veterans General Hospital, one of central Taiwan’s busiest medical facilities.

The simulation tracked minute-by-minute movements of nurses, patients, and robots, identifying workflow bottlenecks and opportunities for automation. By mapping every medication handoff and movement within the ward, Foxconn’s system was able to pinpoint repetitive tasks that could be streamlined, reducing inefficiencies while preserving the human core of caregiving.

Data collected from Nurabot’s onboard sensors and SLAM (Simultaneous Localization and Mapping) navigation technology enabled real-time coordination between staff and machines.

Foxconn’s entry into healthcare AI reflects a longer-term strategic pivot that began in 2019 when the company identified electric vehicles, robotics, and digital health as future-facing pillars. Lin describes the expansion as “a commitment to reshaping how healthcare is delivered and experienced.”

With these efforts now moving from pilot to implementation, Foxconn sees an inflection point on the horizon. Lin projects that the next five years will bring wider adoption of predictive healthcare, remote monitoring, and precision medicine, all underpinned by AI, robotics, and digital twin platforms.

“These advances will alleviate the strain on our healthcare workforce, streamline hospital operations, and most importantly, empower better patient outcomes,” she says.

A company for the AI age

Beyond technological vision, Foxconn’s push into frontier AI is fueled by record-breaking financial performance. In 2024, the company posted revenue of NT$6.86 trillion (approximately US$208 billion), with fourth-quarter revenue rising 15% YoY to NT$2.13 trillion, its highest quarterly result to date.

Foxconn’s gross margin was lifted in Q4 to 6.15% due to revenue growth and improved product mix. Full-year net profit reached NT$152.7 billion (another record), giving the company ample room to fund its AI transformation without sacrificing financial stability.

According to company executives, some of these gains are being driven by AI itself. Foxconn’s Genesis platform, developed with Boston Consulting Group, has reportedly improved productivity by between 30% and 50% in certain factory operations by applying generative AI to optimize workflows and reduce inefficiencies.

Still, Foxconn’s reinvention is far from guaranteed. The company faces stiff competition in every arena it’s entering – from U.S. and Chinese giants in AI research to legacy urban infrastructure players in the smart city space. Scaling up AI platforms in sectors as diverse as manufacturing, mobility, and healthcare will require not only technical muscle, but trust, interoperability, and a willingness to operate in more collaborative ecosystems.

To its advantage, Foxconn brings scale, speed, and institutional expertise to the table. Its decades of experience in system integration, logistics, and global operations give it a foundation few others can claim. With operations across the Americas, Asia, and Europe, innovations developed in a single facility can be rapidly replicated across its network, offering a potential edge in turning pilots into scalable solutions. The company is also expanding its role as an enabler of the broader AI economy through partnerships with American giants like Google, Microsoft, and Nvidia.

Whether this transformation will succeed remains to be seen. But one thing is certain: Foxconn has moved beyond the role of a contract manufacturer to that of a company betting that Taiwan’s legacy of hardware excellence can serve as a launchpad into the next era of intelligent systems. If the gamble pays off, Foxconn won’t merely adapt to the AI age – it will help define it.