GSME builds the chips behind the chips you've heard of.

A back office for the front of the chip industry.

There is a strip of road in north San Jose - Zanker Road, low buildings, no signage worth photographing - that has produced more silicon than several mid-sized countries. GSME's office sits at number 2880. From the outside it looks like any other tenant. Inside, a small army of engineers is running the parts of a chip company most chip companies would rather not run themselves.

The pitch is unglamorous and, for that reason, sticky. Fabless startups, particularly the ones building specialized GPUs, low-power IoT parts, and the wireless modules going into 5G and 6G gear, do not want to staff a tape-out team, negotiate a TSMC shuttle slot, or hire a quality engineer who knows the difference between a Cadence Virtuoso flow and a Synopsys PrimeTime sign-off. They want the silicon. GSME hands them the silicon.

Four numbers that explain GSME more honestly than any About page.

The semiconductor industry forgot how to be small.

The industry's center of gravity is enormous. TSMC is the size of a small economy. A new fab costs $20 billion. A modern design team for a single SoC easily clears a hundred people. And yet most of the interesting silicon being designed right now - the niche RF parts, the custom power management chips, the application-specific accelerators - comes from teams of fifteen to fifty, working on a shoestring, racing a runway.

Those teams have a problem. The infrastructure they need to ship a chip - foundry relationships, EDA licenses, packaging, QA, supply chain - is priced and organized for companies a hundred times their size. Roll your own and you bleed cash. Outsource it piecemeal and you bleed time. Most do both.

GSME was set up to take that second problem off the table. The founders had spent careers inside the back office of large semiconductor companies and watched, slightly bemused, as every small fabless customer reinvented the same operations from scratch. The bet was that those services could be unbundled, sharpened, and sold by the project.

A 28-year veteran and a $2 billion P&L decided to start over.

Farhat Jahangir founded GSME in 2021 after a career that, by industry standards, was already finished. He helped take Quantenna Communications public in 2016. He watched ON Semiconductor acquire it for $1.1 billion in 2019. He stayed on at Onsemi as General Manager and VP of Manufacturing, responsible for a 300mm fab in East Fishkill - the old GlobalFoundries/IBM site - running a $2 billion revenue stream. He has an MSEE from Arkansas and a 2020 induction into the Arkansas Academy of Electrical Engineering. The reasonable next move was a board seat.

He started a services company instead. The wager was modest in scope and unfashionable in shape: combine design, manufacturing, and quality assurance under one roof, sell to fabless and systems companies, and grow by being useful rather than visible.

The unfashionable part has aged well. Five years in, the company has the funding and the customer mix to suggest the model is doing what it was supposed to do.

Five years, four offices, one Muse acquisition, and a Series B that nobody live-tweeted.

Seven services, one back office.

GSME's catalogue reads less like a startup deck and more like a foundry partner's services menu. Which is the point.

Receipts, in three categories.

The customers are listed obliquely on the website and bluntly on the investor materials. High-end GPU companies. Specialized CPU shops, including a public partnership with Andes Technology on RISC-V. Low-power IoT device makers. Wireless product companies, including Tarana Wireless. The partner stack is the standard issue of modern silicon: TSMC for fab, Cadence and Synopsys for EDA, EQSemi in the ecosystem, Muse for shuttles.

Weights inferred from the public Series B announcement. Treat the percentages as directional, not gospel.

The funding round, announced in January 2026 and led exclusively by Maverick Silicon, was the kind of raise that gets a respectful trade-press paragraph and nothing on the consumer wire. $35 million, all into Series B, on top of an earlier $18 million. Maverick's fund is a vertical chip investor; choosing them over a generalist signals that GSME wants money that already understands the patient capital required to build a foundry services business.

Take the hassle out of making a chip.

The official statement is "a value-added semiconductor solutions provider," which is corporate enough to slide off the page. The working translation is more useful: spare customers the cost and the time of building a back office they will never staff well at their scale. Provide design, manufacturing, and QA as a single integrated service. Charge fairly. Show up in the four time zones where modern chips actually get made.

It sounds, viewed from a distance, like the kind of unbundling that has reshaped other industries - the AWS-for-this, the Shopify-for-that. The comparison is mostly wrong. AWS scaled because servers commoditized. Chips do not commoditize. Every node is a fresh fight. What GSME is selling is not standardization but accumulated relationships - foundry slots, packaging vendors, IP libraries, QA engineers - that take years to assemble and minutes to lose.

The middle of the chip stack is the next interesting place to stand.

For a decade the headlines belonged to two ends of the industry: the hyperscalers buying GPUs, and the foundries baking them. The middle - design services, packaging, QA, supply chain - was the part you tolerated. The thesis underneath GSME is that the middle is about to matter more, because the customers in it are about to multiply.

RISC-V is bringing new CPU teams into the world. 5G and 6G are minting new RF startups. IoT is finally past the trough of disillusionment and into the unsexy phase where everyone ships parts. None of those teams can afford a full back office. All of them need one.

That is the bet. It is not a particularly heroic bet. It is, however, a bet that has been right for as long as the industry has existed - that the people who actually ship the silicon will always need someone to run the operations underneath. GSME is making it again, with the patience of people who have already done the loud version once.

Zanker Road, five years in.

The building still does not have a sign worth photographing. The strip mall around it still looks like every other industrial cluster in north San Jose. Inside 2880, the engineers have multiplied, the meeting rooms now include video calls to Hanoi and Hsinchu and Muscat, and the customer list has the kind of names that get redacted in board decks.

If GSME wins, it will not look like a win. There will be no consumer product on a shelf. There will be more chips, in more devices, shipped by more small teams who never had to learn the unloved parts of the trade. The back office will have done its job. Which, in this industry, is the highest compliment available.