Mythic AI has established a Centre of Excellence in Bengaluru, bringing development of its analogue AI accelerator chips directly into India, CRN Asia reports.

The move puts one of the few companies pursuing analogue compute-in-memory architectures for AI inference inside one of the world's largest concentrations of semiconductor design engineers. Bengaluru already hosts design centers for most global chipmakers, and Mythic's decision signals that AI-silicon startups now view Indian engineering capacity as core development muscle rather than a cost-arbitrage afterthought.

Mythic's technology stands apart from the mainstream. Instead of digital SRAM-based accelerators like GPUs, the company processes neural network computations directly in analogue flash memory cells, performing matrix multiplication in the memory array itself. The approach targets edge inference — vision systems, robotics, drones and industrial applications — where power budgets are tight and always-on operation matters more than raw peak throughput.

The Bengaluru Centre of Excellence will host analogue AI chip development work, meaning Indian engineers will contribute directly to the design of future Mythic silicon rather than supporting existing products. That distinction matters for India's semiconductor ambitions: the country has long dominated chip design services, but companies placing next-generation architecture work there are comparatively rare.

The announcement arrives as analogue and near-memory computing gains renewed attention. Digital AI accelerators face a memory-bandwidth wall, and hyperscalers are pouring capital into HBM and advanced packaging to work around it. In-memory analogue compute attacks the problem at the root by eliminating the data movement between memory and processing units, offering power efficiency advantages that conventional architectures cannot easily replicate.

For India, the CoE reinforces a national strategy that has so far emphasized design capability over leading-edge manufacturing. The government's India Semiconductor Mission has attracted fab commitments, but the design-services ecosystem — concentrated in Bengaluru, Hyderabad and Noida — remains the country's strongest lever in the global chip value chain. An analogue AI startup embedding core R&D there strengthens that position.

Mythic's path has not been linear. The company weathered a funding crunch that stalled its first-generation product, then restructured and returned with a second-generation platform aimed at edge AI applications where its analogue architecture delivers the greatest efficiency advantage. Competing at the edge against established players — from NPU-armed smartphone silicon to purpose-built inference accelerators — requires sustained engineering investment, which a larger distributed design organization can support.

The commercial logic of the Bengaluru expansion is straightforward: analogue and mixed-signal chip design is a specialized discipline, and India produces a deep pipeline of engineers in exactly this domain. Widening the talent aperture shortens design cycles and spreads development risk as Mythic iterates its architecture.

CRN Asia's report does not specify headcount targets for the center, the timeline for chips developed there, or the investment value of the facility. The announcement itself — a dedicated analogue AI development site in India by a US-based in-memory computing company — is the substance of the news.

How quickly the Bengaluru team's output translates into shipped Mythic silicon will determine whether this is a talent play or a genuine second design hub. For now, India's role in AI chip development has gained a foothold in one of the industry's most architecturally distinctive camps.