1. Rapidus semiconductor program falls behind on a key 2nm test
Rapidus semiconductor development has hit a delay: the Japanese government-backed chipmaker has not yet demonstrated functioning 2nm SRAM memory cells, a milestone it had aimed to reach by summer 2026, as of mid-November. The shortfall leaves the company several months behind its internal schedule and puts pressure on its plan to start mass production in the second half of fiscal 2027, which ends in March 2028.
The missed checkpoint is significant because of where it sits in the development sequence. A working SRAM cell is one of the first practical proofs that a new manufacturing process can produce reliable circuits, and much of the work that follows depends on it.
Rapidus has not changed its public stance. The company maintains that it is still working to launch volume production in the second half of fiscal 2027, the timeline it has promised from the start.
2. What is Rapidus?
Rapidus is a Japanese chipmaking startup created to bring leading-edge semiconductor manufacturing back to Japan, with heavy financial backing from the national government. Its goal is to produce logic chips on a 2nm-class process, the most advanced generation of manufacturing technology now being developed by the world's top foundries.
A foundry is a company that manufactures chips designed by other firms. Competing at the leading edge requires mastering processes in which transistor features are measured in a handful of nanometers, a level of precision only a few companies in the world have achieved. Rapidus is attempting to join that group from a standing start, which is part of why its milestones attract close attention.
Pilot production takes place at the company's IIM-1 facility. Rapidus keeps the details of its process confidential, so outside observers depend on company statements and on reports from people familiar with its progress to gauge how development is going.
In practical terms, the missed checkpoint means that one of the earliest building blocks of the process has not yet been shown to work as intended, several months after the date the company had planned for it.
3. What a 2nm process means in practice
The label 2nm refers to a generation of chip manufacturing technology rather than a single physical measurement. Each new generation packs more transistors into the same area, which allows chips to run faster, consume less power, or both. Processors for data centers, smartphones and artificial intelligence workloads are among the products that benefit most from moving to the newest generation.
Reaching a new generation is one of the hardest tasks in modern industry. Manufacturers must control hundreds of processing steps with extreme precision, and small variations in any one of them can ruin the structures being built. Even established foundries with decades of experience spend years bringing each new process to the point where it can be sold to customers.
For a startup, the climb is steeper still. Rapidus does not have a previous leading-edge process to build on, so it is learning many of these lessons for the first time while working to a fixed public timetable. That context helps explain why early checkpoints such as SRAM carry so much weight in judging whether the overall plan is realistic.
4. Why a tiny memory cell matters so much
SRAM, or static random-access memory, is a type of very fast memory built directly onto processors, where it serves as cache that keeps frequently used data close to the computing cores. Because SRAM cells are small, dense and repeated millions of times across a chip, they are a demanding test of whether a manufacturing process can produce tiny structures consistently.
For a foundry bringing up a new process, SRAM effectively acts as a proving ground. It is widely treated as the testing foundation for advanced logic manufacturing: if the memory cells do not work reliably, more complex circuits built on the same process are unlikely to work either.
That is why the delay reaches beyond a single test. Until SRAM performs reliably, the later stages of development remain exposed. Those include finalizing the design kits that customers use to plan their chips, producing test chips for those customers and, eventually, reaching stable yields in high-volume production. Each step builds on the one before it, so slippage at the start of the chain tends to push everything after it.
SRAM is also a useful diagnostic. Because the same small cell is repeated so many times, failures show up quickly and in large numbers, giving engineers statistical evidence about which steps in the process are causing problems. A process that cannot yet produce reliable SRAM is therefore still in the stage of identifying and fixing basic manufacturing issues.
5. How Rapidus got here
The company did reach a real milestone earlier in its development. In July 2025, Rapidus produced working prototype transistors using a gate-all-around (GAA) design, the transistor architecture adopted for the most advanced chip generations because it gives better control over current flow than older designs.
The difficulties emerged after that achievement. Making individual prototype transistors work is a different challenge from turning them into a process that can be manufactured at scale. Rapidus has reportedly struggled with integration, connecting transistors to one another and to the other components that make up a functioning chip, and with the slow pace of yield improvement.
Yield, the share of chips on each silicon wafer that come out working, is one of the decisive measures of a manufacturing process. A process can produce impressive individual results in the lab and still be far from commercially viable if too many chips on each wafer fail.
The sequence from prototype to production is often the longest part of bringing up a new process. Early demonstrations show that the basic physics works, but the engineering needed to make that result repeat across entire wafers, day after day, is where most of the time and money is spent.
6. Is a one-year delay coming?
Rapidus is reportedly considering pushing back the start of 2nm mass production by about one year, which would move the launch from the second half of fiscal 2027 into fiscal 2028. The company has not announced any such change and continues to say it is aiming for its original deadline.
The gap between the company's public timeline and the reported internal picture is the core tension in the story. Rapidus says it is making every effort to meet its target, while the missed SRAM checkpoint suggests the schedule has become considerably tighter than the official plan implies.
A one-year slip would matter beyond the company itself. In leading-edge manufacturing, timing determines which customers commit to a process and how a new entrant compares with established rivals, which continue to advance their own technology while a newcomer works through its early problems.
7. Building a customer ecosystem
Rapidus has been preparing the ground for future customers alongside its manufacturing work. The company has signed 17 chip design partners, including the electronic design automation company Synopsys, to help prepare design tools and services for companies that might eventually manufacture at its plant.
Electronic design automation software is what chip designers use to lay out and verify their products before anything is manufactured. For a new foundry, having these tools tuned to its process is essential, because customers need to be confident their designs will translate into working silicon.
That ecosystem work, however, depends on the underlying process maturing on time. Design partners and prospective customers need stable process information to plan around, and the SRAM milestone is one of the first signals that such stability is within reach.
Customer commitments are especially sensitive to timing. Chip designers typically plan products years in advance and need confidence that the manufacturing process they choose will be ready when their designs are. A foundry whose schedule appears uncertain can find it harder to win those early commitments, even if its technology eventually performs well.
8. Public money and Japan's chip ambitions
The Japanese government has committed more than ¥2 trillion to Rapidus. That backing reflects Tokyo's aim of rebuilding a domestic capability in advanced logic chips, an area where Japan has lagged for years.
With that much public money involved, technical progress at Rapidus carries more than commercial weight. A delay in development can quickly turn into a question about funding and about whether the national strategy for semiconductors is on track. The stakes help explain why every milestone, met or missed, is closely watched.
9. What to watch next for Rapidus
The clearest signal ahead is whether, and when, Rapidus announces that it has produced working 2nm SRAM cells. That confirmation would indicate the process has cleared one of its most important early hurdles and would give its design partners and potential customers firmer ground to plan on.
Each month without that confirmation shortens the time left for the stages that follow, from customer design kits to test chips and volume yields, before the company's stated production window in the second half of fiscal 2027. Any formal decision to move mass production into fiscal 2028 would be the most direct sign that the schedule has changed.
Until then, the company's public position and the reported state of its development point in different directions. The next update on SRAM will likely say more about the realistic timeline for Japan's 2nm ambitions than any official reassurance.
The outcome also matters for Japan's broader industrial strategy. A successful 2nm launch would give the country a domestic source of the most advanced logic chips, while a prolonged delay would extend its dependence on overseas suppliers for that technology.
