In September of last year, reports began to surface that microprocessor giant Intel (INTC 4.88%) had purchased a small chip technology start-up known as Soft Machines for approximately $250 million.
Soft Machines essentially claimed that it was working on microprocessors that could deliver both much higher performance and power efficiency than what typical processor designs could.
Soft Machines' marketing claims of the performance of its chip designs. Image source: Soft Machines.
The Register, which first reported that this acquisition had happened, said that the company's chip designs "have not performed as well as promised," however.
Intel didn't publicly announce this acquisition at the time, but over time it became increasingly simple to verify that the transaction did occur (I monitored the LinkedIn profiles of several Soft Machines employees and saw many of them update their profiles to show that they now worked at Intel).
At its Feb. 9 investor meeting, Intel finally publicly announced that it had purchased the company and provided a little bit of its reasoning behind the buy. Let's dig in.
Giving its CPU tech a boost
CFO Bob Swan said that during 2016, Intel acquired several relatively small companies, one of which was Soft Machines. According to Swan, the rationale behind these acquisitions was to "accelerate [Intel's] technologies and [its] time to market in critical areas that we think are big differentiators for one of our businesses, if not all."
On Soft Machines specifically, Swan explained that "this transaction really gives [Intel] a team of skilled engineers to further enhance [its] CPU-oriented products."
Image source: Intel.
Given that virtually all of Intel's revenue comes from "CPU-oriented products," this is kind of a big deal and probably merited some additional background or context.
I'd like to offer up a potential hypothesis for why Intel might have bought Soft Machines.
Intel may be seeding third processor core team
It is well-known that Intel has two major processor core development teams -- one in Hillsboro, Oregon, and one in Haifa, Israel.
In the past, Intel's processor development methodology was known as "tick-tock." In "tick" years, Intel would take a previous processor architecture, sometimes make a few tweaks along the way, and implement it in a brand-new chip manufacturing technology.
In "tock" years, Intel would introduce a significantly enhanced/brand new processor architecture implemented in a proven manufacturing technology.
Under Intel's current product development methodology, it will use the same basic manufacturing technology (with performance enhancements) for three to four generations.
What's interesting is that Intel has also used the same processor core architecture (called Skylake) for two generations in a row now (sixth-generation Core processor family as well as the seventh-generation Core processor family) and is expected to use that same Skylake architecture for a third generation.
Going forward, Intel could be interested in bringing more substantial architectural changes each year to deliver more compelling products and to strengthen its competitive positioning. To speed up that pace of innovation, the chipmaker may have bought Soft Machines to "seed" a third major processor development team.
Indeed, India-based publication The Economic Times quoted someone with knowledge of the situation as having said that "the plan is to see that the team size is increased more than 10-fold over the next two-three years to nearly 2,000."
This doesn't smell like a "tuck-in" technology acquisition designed to augment current efforts -- it looks to me that Intel had planned to put together a third processor development team anyway and figured that buying the Soft Machines talent would be a good way to get that going.
