At a macro level, the merger is about more efficiently marrying analog with digital, enabling new markets for the merged company and new applications for customers.
At a tactical level, it highlights an emerging battle between programmable logic vendors and microcontroller companies. Indeed, it sets the stage for what will be a thoroughly enjoyable smackdown between Actel’s Fusion and SmartFusion product lines and Cypress’s PSOC platform.
Are you, as the man says, ready to rumble?
At a financial level, MicroSemi clearly needs to move up the value chain. Its stock has, for the most part, under-performed its peer group since the industry’s public companies began to turn northward in mid-November 2008.
Selling onesie/twosie analog parts is a death dance with cost-cutting and volume-stalking, so most analog vendors are feverishly doing what their digital cousins did decades ago: integrating what makes sense into modules and selling the cost-saving, real-estate-maximixing, power-reducing benefits--see
You get the picture. And more is better in a commodity market.
So to synergy: Mil-aero and rad-hard are not sexy, but it's a solid business, and both Microsemi and Actel have excellent positions there. Check.
Longer term, though, this merger reflects the new semiconductor industry and the long-simmering trends that have gotten us here.
As FPGAs slowly overcome their historical cost issues as process geometries get smaller, they become more and more popular. More end markets have shorter half-lives, which makes FPGAs even more appealing because they don’t have to worry about the dreaded volume cross-over where ASICs win the day on cost.
The historic evolution of FPGAs has been to slowly add IP outside the core fabric, starting first with processor cores. Actel carved a niche for itself with Fusion, and this year, with the rollout of SmartFusion (FPGA+CPU+analog features), took analog functionality to a different plane.
Actel doesn’t look for Altera or Xilinx on this field of battle; they look to controller vendors and especially Cypress and its programmable system-on-chip (PSOC).
They consider Cypress to be a little stronger right now on the analog side, while they consider their own architecture to have much better programmability features.
Here’s a list of analog peripherals in the Cypress CY8C55:
1.024 V±0.1% internal voltage reference across –40 °C to 85 °C (14 ppm/°C)
Configurable delta-sigma ADC with 12- to 20-bit resolution
Two SAR ADCs, each 12-bit at 1 Msps
80 MHz, 24-bit fixed point digital filter block (DFB) to implement FIR and IIR filters
Four 8-bit 8 Msps IDACs or 1 Msps VDACs
Four comparators with 75-ns response time
Four uncommitted opamps with 25-mA drive capability
Four configurable multifunction analog blocks.
The Actel SmartFusion devices offer analog features such as:
DACs (12-bit sigma-delta)
bipolar high voltage monitors.
These are not apples-to-apples comparisons, but you get the drift. These features were unheard of in FPGAs a few years ago.
System integration story
So there is a device-integration play at work here, but there's also a system-integration story as well.
If Actel is chasing an motor control win, don’t you think they love the idea that they can say “oh, by the way, since you're buying our FPGA, we have the peripherals you need too, including transformers, drivers, IGBTs, regulators, and more.”
You bet they do.
The same works for the Microsemi sales force from the analog perspective.
The bottom line is the semiconductor industry is reforming itself yet again. Passing are the days when everyone specialized in something vertical. Horizontal integration is the current wave, and it’s what drives mergers like this.
There's a lot here to digest. But on the whole it makes sense. The only disagreement I might have is that mil/aero and rad-hard are NOT "sexy". Designing for military and the medical applications is challenging: the systems need to last and be secure. What's more, reliability and security is starting to be taken seriously in other apps as well. On the whole, I think Actel was smart selling and letting the other two main FPGA behemoths duke it out for the rest of the market pie. The only question is did Microsemi give Actel's shareholders the highest value. Already two class action suits are claiming Actel was not shopped around enough for Actel shareholders to get their money's worth.
Great article Brian, and I completely agree this is an example of great synergy between two well-aligned semi companies. There's going to be a lot of activity in this space over the next couple of years so you are going to have your hands full covering them! My only counterpoint: Mil-aero and rad-hard are going to be a bigger market than industry watchers are currently anticipating. Smart systems such as drones and hardened systems for gas line monitoring and control, etc., are going to require just this kind of electronic content. Should be interesting!
It will be interesting to see how those suits shake out, and it's probably hard to prove malfeasance in this case. It's been a rough decade for ACTL shareholders, but then again the company's stock performance has been in the middle of the pack vis a vis other public programmable logic companies during that time.
Microsemi's long term and continuing mistreatment of MIL/Aero customers is legendary. Microsemi's capabilities lie largely in producing datasheets, but not parts. Not on time.
Heaven help current Actel users. Stand by to be stabbed in your nich.
I love the idea of the combined analog and digital programmable devices! I can see a number of uses for this hybrid design capacity. In looking at the older traditional approach, mix of analog and digital chips, there was the cost/size/complexity issues that slowed time to market. With a combined single chip solution that can be reprogrammed on the fly, the designer now can quickly code, test, and implement with the basic evaluation boards offered and speed to a final solution. I look forward to the future offerings and increased capacities of these devices.
I also like the idea to put programmable logic together with mixed signal product, provided it won't add noise to the high performance analog stuff. Anyway, this is a good direction to go and I hope to see more in the future.