Hi, guys i'm new to this industry and am having difficulty identifying the weakness of being an IC design service provider company. I have just joined an IC design service company. As a prerequisite to securing the position as a key account manager in this company is to write a business report, and im finding it difficult to identify the weaknesses of being an "IC Design company". It's not so difficult to find the weaknesses of Integrated circuits as a whole but when it comes to just the design, Google is less than helpful. Can anyone here please help me? Any help is much appreciated. Thanks

 

Welcome to AAC!
I would think one weakness is that you are in competition with the in-house design teams of the big manufacturers of ICs. They have insider knowledge of proprietary manufacturing processes.

 

Welcome to AAC!
I would think one weakness is that you are in competition with the in-house design teams of the big manufacturers of ICs. They have insider knowledge of proprietary manufacturing processes.

It's a mixed bag in that regard. If you want to compete with the big houses on the types of designs that they specialize in, then you are pretty much toast. But there are plenty of types of designs that they don't specialize in and that they are ill-prepared to compete with a small, nimble company. The big companies focus on optimizing their processes to be really good at the collection of designs that they have chosen to emphasize. The people they hire, the tools they use, the workflows they adopt, how they document things, you name it. They tend to become too rigid and their very sophistication works against them when they step out of their wheelhouse. The small (4-12 engineers), company I worked for did designs for some of the big houses, including Texas Instruments, at a fraction of the cost and time that they had sometimes spent years and millions of dollars working on. But these were lunatic fringe designs that we specialized in, but that's because we had chosen to focus on those kinds of designs using small teams (one or two engineers per design), cheap PC-based tools, and very manually-intensive workflows (the engineer gets to layout out every transistor and track, which is why our designs tended to be array-oriented).

 

Welcome to AAC!
I would think one weakness is that you are in competition with the in-house design teams of the big manufacturers of ICs. They have insider knowledge of proprietary manufacturing processes.

Thanks very helpful feedback.

 

It's a mixed bag in that regard. If you want to compete with the big houses on the types of designs that they specialize in, then you are pretty much toast. But there are plenty of types of designs that they don't specialize in and that they are ill-prepared to compete with a small, nimble company. The big companies focus on optimizing their processes to be really good at the collection of designs that they have chosen to emphasize. The people they hire, the tools they use, the workflows they adopt, how they document things, you name it. They tend to become too rigid and their very sophistication works against them when they step out of their wheelhouse. The small (4-12 engineers), company I worked for did designs for some of the big houses, including Texas Instruments, at a fraction of the cost and time that they had sometimes spent years and millions of dollars working on. But these were lunatic fringe designs that we specialized in, but that's because we had chosen to focus on those kinds of designs using small teams (one or two engineers per design), cheap PC-based tools, and very manually-intensive workflows (the engineer gets to layout out every transistor and track, which is why our designs tended to be array-oriented).

Thanks for your input. Found your explanation very informative and easy to understand.

 

It's a mixed bag in that regard. If you want to compete with the big houses on the types of designs that they specialize in, then you are pretty much toast. But there are plenty of types of designs that they don't specialize in and that they are ill-prepared to compete with a small, nimble company. The big companies focus on optimizing their processes to be really good at the collection of designs that they have chosen to emphasize. The people they hire, the tools they use, the workflows they adopt, how they document things, you name it. They tend to become too rigid and their very sophistication works against them when they step out of their wheelhouse. The small (4-12 engineers), company I worked for did designs for some of the big houses, including Texas Instruments, at a fraction of the cost and time that they had sometimes spent years and millions of dollars working on. But these were lunatic fringe designs that we specialized in, but that's because we had chosen to focus on those kinds of designs using small teams (one or two engineers per design), cheap PC-based tools, and very manually-intensive workflows (the engineer gets to layout out every transistor and track, which is why our designs tended to be array-oriented).

When you mentioned you have previously done work for Texas Instruments, does that mean that your working in Malaysia? Cause I'm in malaysia and completely new to the IC industry. I have experience in the sales line but nothing in the IC industry. Any advice you can give to someone newly joining the industry? I will be joining as a key accounts manager (glorified saleman).

 

When you mentioned you have previously done work for Texas Instruments, does that mean that your working in Malaysia? Cause I'm in malaysia and completely new to the IC industry. I have experience in the sales line but nothing in the IC industry. Any advice you can give to someone newly joining the industry? I will be joining as a key accounts manager (glorified saleman).

Nope. We are in Colorado. But virtually none of our customers were. We did designs for TI, Sony, Hughes, Ball, JPL, and many other big names -- though most of our work was for universities and startup companies -- often without ever physically meeting our customers. Two or three conference calls, a good quantity of one-on-one phone calls, and lots and lots of e-mails.

 

As a prerequisite to securing the position as a key account manager in this company is to write a business report, and im finding it difficult to identify the weaknesses of being an "IC Design company".

If by "IC Design company", you mean fabless; if you work on "large" designs, you'll never realize the cost margins and advantages of being able to design on a leading edge process and co-develop design rules the way that design teams of an IDM (Integrated Device Manager) can.

I worked in design teams a large IDM and we could negotiate with process development on design rules that would allow us to improve density/die size (we were the first high volume microprocessor on the new process; after our project, the rules were frozen). If we had questions on the intent of a certain rule, we had access to the process developers.

1. When using foundry services, you're unlikely to have access to all process capabilities.
2. Your manufacturing costs will always be higher than an IDM.
3. You'll never have access to bleeding edge technology; that is, be the first design on a process.
4. You'll have no control over fab priorities.

I knew someone who worked at a company that was designing a knock off a microprocessor 20 years ago. They weren't doing a full custom design, at least in the parts where it would make a difference in area and performance, because they didn't have the resources (people or tools) and didn't have access to a "good" process technology. Their design was so large that it had to be put in two packages. As I recall, it was never a success. But they did the design with a handful of designers, probably 5% of what my company would have used, and did the design much faster.

 

If by "IC Design company", you mean fabless; if you work on "large" designs, you'll never realize the cost margins and advantages of being able to design on a leading edge process and co-develop design rules the way that design teams of an IDM (Integrated Device Manager) can.

I worked in design teams a large IDM and we could negotiate with process development on design rules that would allow us to improve density/die size (we were the first high volume microprocessor on the new process; after our project, the rules were frozen). If we had questions on the intent of a certain rule, we had access to the process developers.

1. When using foundry services, you're unlikely to have access to all process capabilities.
2. Your manufacturing costs will always be higher than an IDM.
3. You'll never have access to bleeding edge technology; that is, be the first design on a process.
4. You'll have no control over fab priorities.

I knew someone who worked at a company that was designing a knock off a microprocessor 20 years ago. They weren't doing a full custom design, at least in the parts where it would make a difference in area and performance, because they didn't have the resources (people or tools) and didn't have access to a "good" process technology. Their design was so large that it had to be put in two packages. As I recall, it was never a success. But they did the design with a handful of designers, probably 5% of what my company would have used, and did the design much faster.

As a general rule, I think everything you say is usually the case. There are exceptions, as there almost always are. The small company I worked for helped with several aspects of process and design rule development for a few processes on TSMC, Tower, and a couple of others. Our designs were the first design on a couple of those processes. But it was an odd situation in that our design happened to be a match for what they were trying to achieve with their new process and so we were a good guinea pig for them. It was mutually beneficial. Our customer got the fab for next-to-nothing but had to be patient through a few failed fabs while they worked out the kinks exposed by the pressure our design placed on their process. Since no one else could come close to fabbing imaging chips the physical size that the customer needed, they had little choice but to deal with the delay and almost certainly ended up with working chips a couple years before they probably would have otherwise.

Back when I first started there, in 1995, our lunatic fringe designs almost always involved intentionally violating the design rules, sometime flagrantly. Our president had a couple decades of experience working at a large company with in-house fab facilities and worked directly with the fab folks in the way you describe and so knew what could and couldn't be pushed (and a good feel for how far they could be pushed). That all ended at about the 250 nm node when we discovered, the hard way, that violating design rules was simply no longer a viable design strategy.

 

Since no one else could come close to fabbing imaging chips the physical size that the customer needed, they had little choice but to deal with the delay and almost certainly ended up with working chips a couple years before they probably would have otherwise.

I think Altera experienced something similar when they signed up as an Intel foundry customer. They had to deal with problems and delays in trying to ramp up on a process that was even giving internal design teams problems. But the payoff was that their main competitor was unable to compete because there were no other comparable foundry processes.