A Brief History of Kilopass

Kilopass was founded back in 2001 by Jack Peng, whose background was in FPGAs with his most recent position being manager of technology development at Actel (now part of Microsemi). The idea was to build a company making one-time-programmable (OTP) memories using anti-fuse technology. Fuses in home-wiring (OK, I know, we all have circuit-breakers these days) work by melting metal and thus breaking the circuit. Anti-fuses work by punching a hole in the gate-oxide using a high voltage and thus making the circuit. Anti-fuses had been around for a long time. In fact I remember in 1988 at VLSI Technology when we started building anti-fuse devices for Quicklogic. Actel also had anti-fuse FPGAs.

The problem with the old approach to anti-fuse is that the comparatively thick gate-oxide and the high voltage required for programming necessitated a non-standard process. What Dr Peng realized was that at 180nm it would be possible to build anti-fuse devices on standard CMOS logic since the gate-oxide breakdown voltage (for programming) was less than the junction breakdown. The higher voltage needed for programming could be generated entirely on-chip.

In the early days of Kilopass, one of the challenges was that customers and foundries were not completely convinced that this approach was not going to cause reliability issues. It really took about 5 years before they got strong customer traction.

The basic idea has not changed since founding, although of course the memory cells have got smaller as they have ridden the technology wave down through process nodes as you would expect. Kilopass is an IP company. They do not manufacture OTP memories but rather they license their technology to companies to incorporate OTP registers or memories onto their SoCs.

I talked to Linh Hong, who is one of the longest serving employees at Kilopass and currently VP sales. She joined as the company in 2006 as it closed its series-C funding and was starting to achieve design-ins. Eight years ago they had around 30 customers, now they have somewhere between 150 and 200.

Today Kilopass is still an IP company supplying OTP memories as both sizable memories and as small register files. It has foundry support at TSMC from 180nm with 16nm in development. SMIC from 180nm with 20nm in development. GF from 130nm down to 14nm in development. UMC from 130nm with 28nm in development. And processes from Jazz, IBM, Grace and Dongbu Hitek. Full process grid is here.

Looking on the net I came across a quote from Charles Ng, who coincidentally used to work with me me at VLSI/Compass. He was the VP sales and marketing for Kilopass in its early days. He described their business back then:Kilopass is an IP Supplier of field programmable memory. You can program the memory content. It is an IP embedded in a SoC. Non-volatile memory means that data content is kept after the power is off. Our claim to fame is that we require only standard logic CMOS process without any additional processing at all. Volatile memory requires a complicated process which is very expensive to develop and to use.

The only thing that you might add today is to emphasize that in the current security environment, anti-fuse has another great attribute. It is basically impossible to “read” the value programmed into the OTP memory even with scanning-electron-microscopes and similar tools.

But today, when venture capitalists do little but talk about “pivoting” and “getting to plan B”, it is interesting to see just how a company can successfully keep the same value proposition for over a decade and build it up into a sizable business.

There is a page on the Kilopass website that lists the company’s chronology here.

Kilopass are having an open house on Thursday April 2nd from 11.45am to 2pm (so there is such a thing as a free lunch) at their new offices at 2895 Zanker Road, San Jose. If you will be attending then RSVP here.