What began as a question about what "Made in India" could mean in deeptech has become a company with ambitions that reach far beyond Bihar. Swadeza, founded by Shweta Suman, is now being built from a state more often associated with migration than with semiconductor innovation, and it is using one of India's most visible chip industry stages to make its case.
The company's appearance in the orbit of SEMICON India underscores a larger shift in the country's technology landscape: the idea that advanced engineering companies can emerge from places not traditionally linked to frontier hardware or enterprise software. For Suman, that shift is personal as well as strategic. Her path has moved across borders and back again — school in Oman, engineering at BIT Mesra, a return to Oman, and then a move back to Bihar after marriage. Rather than treat the move as a detour, she turned it into the foundation for a startup.
That decision carries symbolic weight in Bihar, where the dominant story has long been one of talent leaving in search of opportunity. Suman's trajectory runs in the opposite direction. She came back, built a company, entered India's startup ecosystem and is now trying to place a Bihar-based venture on a national semiconductor map.
Swadeza is not a single-product company. It operates across semiconductor technology, IT consulting and development, and 360-degree digital marketing. But its most ambitious work is increasingly centered on AI-native semiconductor engineering. The company says it is building tools for secure, on-premises deployment, a point that matters in an industry where design data is sensitive and cloud-first workflows are not always acceptable.
At the heart of that effort is FORGE, Swadeza's semiconductor platform. The company argues that the hardest part of chip development is not just design, but proving that a design behaves exactly as specified before fabrication. Verification is one of the most time-consuming stages in the chip lifecycle, and mistakes discovered after tape-out can be extraordinarily expensive.
Swadeza's critique of existing AI tools is that general-purpose assistants break down at semiconductor scale. As a design grows, the prompt becomes crowded with unrelated and conflicting fragments, and answer quality deteriorates. FORGE is designed to avoid that problem by keeping the chip in a cited map, sending a language model only the information relevant to a given question, and converting what it learns into deterministic models. The company says the goal is to make each run faster and cheaper than the last.
The platform is organized around four products spanning the chip lifecycle, from verification and physical design to post-silicon validation and standard cells. Its verification tool, DV-FORGE, reads a chip's reference manual into a map and generates the verification plan, coverage model and UVM testbench from it. Swadeza says DV-FORGE has been used on OpenTitan-class SoCs, including signing off OpenTitan Earlgrey, a taped-out root-of-trust chip, in 30 days and finding seven bugs across roughly 40 IP blocks and 2,581 registers. The company says its derivative, Darjeeling, was signed off in 21 days, with most of that time spent in regression.
Swadeza also claims the system uses 80% fewer tokens than an AI assistant and says it has run the pipeline on reference manuals from NXP, STMicroelectronics and Renesas. Those are significant claims in a sector where speed, reproducibility and trust are everything, and where even incremental gains can translate into major savings for engineering teams.
The company's broader pitch is that India's semiconductor ambitions will not be built only by fabs, policy announcements or large industrial players. They will also depend on the software layer — the tools that help engineers verify, validate and ship chips faster. Swadeza is trying to occupy that layer while also making a statement about where innovation can come from.
For Suman, the story is as much about geography as it is about technology. Swadeza's rise from Bihar to SEMICON India reflects a bet that deeptech does not need to be confined to established hubs, and that globally relevant products can be built from places the industry has too often overlooked. Whether the company can convert its technical claims into durable market traction remains to be seen, but its message is already clear: the next wave of Indian technology may not just come from the usual centers of power. It may come from the margins, and it may come with a semiconductor blueprint attached.