Lenovo on Tuesday announced high-performance AI server manufacturing based out of India and a new R&D lab in Bengaluru. The company, which established its Indian operations nearly two decades ago and its Motorola R&D lab a decade ago, is now expanding its India investment with a multi-million-dollar plan. The Puducherry facility will produce 50,000 enterprise AI servers and 2,400 high-end GPU units annually, addressing growing AI infrastructure needs and targeting both domestic and international markets, with over 60 per cent of production aimed at exports across Asia-Pacific. The company's APAC and India leadership spoke with BW Businessworld on Lenovo's manufacturing capabilities based out of Puducherry facility and India. Excerpts:
Lenovo has been manufacturing in India since 2005. What changes have influenced Lenovo’s manufacturing approach, particularly with the new AI angle, and what was the leadership’s thought process behind these shifts?
Amar Babu, President of Lenovo Asia Pacific: We have been manufacturing in India for two decades. So, manufacturing in India is not new to us and over a period of time we have expanded our capabilities here. We used to manufacture PCs, then we started manufacturing tablets, and we started manufacturing smartphones. It has been a continued evolution that whenever there is the right business model, opportunity and a business case, we would like to manufacture in India because it brings us closer to the customer and also allows us to leverage the local skillset, talent and government incentives and programmes.
Now, we are bringing AI server manufacturing as well. This means we are literally bringing the whole full stack of the portfolio that includes products and services, which is happening out of India, for India, and also for other parts of the world, especially Asia Pacific. India is a key hub for Lenovo worldwide and this is the next step.
Could you give us an idea of manufacturing capabilities of Lenovo currently based out of Puducherry?
Amit Luthra, Managing Director, Lenovo ISG India: Over the years, we have grown both organically and inorganically, acquiring IBM’s server business and Motorola. Our Puducherry facility, with two decades of experience, initially produced PCs and phones, and today, it manufactures Make-in-India PCs, even from the motherboard chipset level. This facility is one of our key manufacturing hubs among 30+ globally, with optimised supply chains and IT systems.
When we decided to expand manufacturing in India, Puducherry was the ideal choice due to its established systems, processes, and focus on supply chain reliability, which is crucial for our servers’ reputation. We have ramped up production, adding assembly lines to manufacture mainstream servers, including 1U and 2U two-processor servers (Intel and AMD), as well as AI-specific 8-GPU machines. This enables us to meet India’s growing demand while also serving global markets.
You have clearly defined that the ambition is of making 50,000 enterprise AI rack servers annually and 2,400 high-end GPU units is not limited to serve the local demand but also serve overseas market. There is an evident supply gap in both AI servers and GPU units globally, but what are your expectations? Who’s placing the order?
Amar Babu, President of Lenovo Asia Pacific: We have already begun and have customers. We are building the infrastructure for AI servers and GPUs, both for India and exports. With the global demand for AI servers, we don’t foresee any issues in utilising this capacity. While ramping up manufacturing takes time to fine-tune processes like testing and engineering, we expect the first units to roll out by late this financial year, around March or early next financial year. From April, we hope, production will ramp up to full capacity.
“We began investing in AI infrastructure in 2017 with USD 1.2 billion and added another USD 1 billion last year, with USD 900 million focused on building platforms, labs, and ecosystems, and USD 100 million dedicated to our AI Innovators Program. This program supports smaller companies, helping them test and scale their technology globally by providing the necessary infrastructure. So far, we have brought in over 60 innovators, resulting in 165+ solutions. This journey, starting with large LLM players, is now expanding as organisations across regions, including India, develop use cases for AI" – Sumir Bhatia, President, Lenovo ISG Group - APAC
What’s Lenovo bringing to the table with the new GPU- and AI server-making capability based out of India in terms of supply chain and server market share?
Amit Luthra, Managing Director, Lenovo ISG India: The opportunity in India is immense. If you look at my server market share report from IDC, the AI server market is adding as much volume in a quarter as the entire traditional server market. This trend has continued through recent quarters, and the AI market is expected to grow significantly, with projections of a 35 per cent CAGR by 2027, possibly even 40 per cent. The key is optimising supply chains, where we leverage our global expertise. Gartner has ranked us 8th in supply chain management consecutively, which ensures fast product availability. And imagine, we are importing 300-kg AI servers from our Mexico factory, which must be transported as dangerous goods, requiring specialised cargo flights. This classification means they can’t be shipped on regular aircraft, resulting in an 8–10-day transit time. Once we start manufacturing these servers in India, the time-to-market will be significantly reduced.
What sort of market share and revenue gains are you anticipating out of this specific manufacturing move based out of India?
Amit Luthra, Managing Director, Lenovo ISG India: Our capacity for now includes 50,000 mainstream servers and 2,400 high-end 8-GPU servers. Imagine the revenue from an 8-GPU server, which typically costs upwards of USD 300,000. The annual revenue generated from manufacturing these high-end systems will be substantial. We are also market leaders in tower servers with a 30 per cent share. For mainstream, enterprise, and hyperscaler workloads, all can be met through infrastructure built in India.
