The semiconductor industry is generating almost as much Fomo as it does revenue, creating a surge of competition for a finite number of fabrication facilities. Thousands of artificial intelligence chip startups across Asia have emerged, many founded by veterans of older companies like Intel and AMD. These firms design chips but rely on specialist foundries for manufacturing, creating a bottleneck that rivals the market dominance of Nvidia.
Access to advanced manufacturing is a major hurdle
For a startup trying to build at the frontier, a relationship with a major foundry can matter as much as the chip itself. Chan Yip Pang, an executive director at Vertex Ventures, notes that TSMC remains the market leader in advanced nodes. Getting an allocation at TSMC is typically not so easy, and the gap between a 3 nm node and a 15 nm node is significant; the former fits transistors closer together, which usually makes a chip faster and more power-efficient, but also much harder and more expensive to manufacture.
Despite this bottleneck, investors continue to back AI chip startups. South Korea’s FuriosaAI is one of the few startups in Asia to have secured production for its advanced AI chips. The entire process was “very complicated,” requiring the startup to obtain TSMC capacity and gain access to high-bandwidth memory as well as packaging partners and customer validation all at the same time. The company has raised US$246 million and shipped products to customers and partners including Samsung and LG. Its flagship AI inference chip, called RNGD, entered mass production in January 2026 on TSMC’s 5 nm process, with an estimated cost of US$10,000 per chip.
Being at the leading edge nodes presents a unique set of challenges for companies like Bengaluru-based Agrani Labs. The company has raised US$8 million from Peak XV Partners and is reportedly in talks to raise more than US$100 million. While investors tell Tech in Asia that CEO Dheemanth Nagaraj and his team are among the strongest chip design talent in the region, Arjun Rao of Speciale Invest notes that even a small tape-out and validated initial volume cost a lot. “Capacity access is a barrier for young startups given the large AI compute build out globally,” he says.
Tape-out is when a finished chip design is sent to the fabrication facility for manufacturing.
Finding a path forward through partnerships
Startups cannot solve the memory constraint alone, and many have been seeking to build partnerships with suppliers. SK Hynix, Samsung, and Micron dominate the market for high-bandwidth memory, and access is vital. “HBM is so important,” says Furiosa’s senior vice-president Alex Liu. “Those are the conditions that you need to win,” though he says it is not about how much money you have. SK Hynix makes it very clear it will not just support any startup.
Related: Tropical Cyclone Names Include Dolphins Dim Sum
Singapore-based Silicon Box aims to ease this obstacle using advanced panel-level packaging to connect the AI processor, high-bandwidth memory and other components into a single chip package. Before engaging with AI-chip startups, Mike Han, chief revenue officer of Silicon Box, says they need to fulfill several criteria. According to Han, Silicon Box assesses the technical capability of a team first and then the track record of the founding team. “Whether they’ve been able to be successful in the past is a key factor,” he notes.
For Furiosa, courting customers did not begin after the engineering was complete; it began alongside the design process. “When you start to design the chip, you need to start to talk to the potential customers to understand the market trends,” says Liu. These conversations went through multiple rounds as the company moved toward production. The company argues that cost and sovereignty issues are among the most important factors for potential buyers. “As Furiosa, we are non-US, non-China technology, so we are neutral to everyone,” he says. “That reduces a lot of the political risk as well.”
Since chips have become central to geopolitical competition, governments are increasingly seeking their own alternatives to US and Chinese suppliers. Rao says that India should have a few strategic AI chip alternatives, not just one. “We’re building data centres, we’ll need the compute, so having homegrown alternatives is a good idea in the medium to long term.” In South Korea, government support has complemented an established memory and semiconductor ecosystem to help produce companies such as Furiosa and Rebellions.
AI chip startups in India and South-east Asia, many of whom have been boxed out of the competition with government-backed entities, are taking a less capital-intensive route. As they struggle to access high-end fabs, many are focusing on designing application-specific integrated circuits, or ASICs, for edge AI or particular workloads. Edge AI chips can power real-time tasks for drones, robots, CCTV cameras, industrial machines, EVs and autonomous vehicles without sending data to a distant data centre. Such chips may not need the most advanced manufacturing capacity or the same volumes of scarce high-bandwidth memory as a frontier AI data-centre chip.
Singapore-headquartered OptoML is building such chips for CCTV cameras, drones and robots. The company is currently piloting its chip with manufacturers, including Chennai-based Murugappa Group and Wheels India. OptoML’s current chips use a 12 nm process node, an older manufacturing technology that is generally cheaper and easier to access than the 5 nm process that Furiosa uses. Startups working with older chip generations have more manufacturing options, such as Silterra in Malaysia, a foundry that can handle more mature nodes.