You are here

Interview with, part three: T.J. Rodgers on Capital Allocation | Cypress Semiconductor

Interview with, part three: T.J. Rodgers on Capital Allocation

Last Updated: 
May 24, 2012


A key activity in any business is capital allocation. In this interview, T.J. Rodgers, CEO and founder of Cypress Semiconductor Corporation, explains how his allocation methods have evolved over the last decade.


DC- You operate in a capital intensive business. How do you go about making those difficult decisions on headcount and allocating capital?

TJR- In No Excuses Management, I wrote about our processes in the chapter People and Capital: How to do More with Less. Those processes have changed. It's a question of scale. The capital and people systems described in the book are right and practicable but they were not scalable for me. However, the mechanisms described in the book are used at the bottom of the company and I manage results at a higher level.

DC- What is the fundamental problem of capital allocation from a CEO's point of view ?

TJR- Every manager wants more machines and more headcount. If I met with each vice president one-on-one they would each explain that if we don’t get this machine or this new hire our productivity will flatten out, we won’t be able to increase productivity yada, yada, yada. At the end of fourteen one on one meetings, I would be convinced to spend more money than we could afford. The result of fourteen seemingly good decisions would be one very bad decision.

DC- So how did you address that in the early days of Cypress?

TJR- When Cypress was small enough, we had a monthly capital meeting where I allowed for a huge amount of input. Each VP would have a wish list and collectively those wish lists were larger than our budget. However, everybody would agree we needed to cut down the list because the VPs knew there was a limit on how much we could spend and how many people we could hire.

Capital allocation issues involved decisions such as choosing between a new machine in the fabrication plant or a better computer to keep financial records. By having all the vice presidents in the room, I could have an inter-disciplinary discussion and make the right trade offs between groups. There was a lot of input, a lot of interaction, and the vice presidents would scratch stuff off the list if a better idea came along or substitute a cheaper way of getting something done. I would only have to make a few decisions. We'd end up with a consensus budget.

As the CEO, I could have made those decisions myself but I don’t trust my own evaluation as much as I trust group consensus. At the end, everyone might not be happy but they feel we've gone through a reasonable process to allocate money.

This method worked in a small organization with a few hundred employees. The vice presidents knew what was going on at the grass roots level and collectively we could make an optimum decision about the best way to spend money.

DC- What happened when the company got bigger?

TJR- By the late 1990s, when the company revenue was around $500 million, the system was clearly broken.

We'd have 14 vice presidents in the room, each of them with 75 line items in the budget. The vice president was no longer capable of telling you in detail what their 75 line items were, so they each brought 15 middle managers who did know what was going on.

When you have a twenty minute debate on whether or not you need a new mainframe computer for the Oracle database, the guys in the assembly area, who just want another piece of wire binding equipment, fall asleep. At this scale, people are only interested in a fraction of the meeting so they sit for three hours to give only ten minutes of input.

DC- So how does the process work now?

TJR- First, we set the overall budget. It's usually pretty straightforward. The CFO can come in and tell you how many dollars for capital and how many dollars for people you can afford as a corporation.

It's important to make that decision first since the vice presidents are more likely to agree to lower numbers—fewer people and less dollars—when they're asked to be surrogate CEOs before they really knew what that budget meant for them. So you put the lid on and get everyone to agree that this is the right thing to do.

Then you have the debate about how to deploy the money to the various groups at a budget level, but not in detail. We don't debate my machine A versus your machine B. We argue on the basis of how things affect the productivity of the organization. For example, one VP will say, "I'm trying to hire people to make chips and that's what we do as a business and you are talking about hiring 10 accountants to close your books faster, that's bullshit." These debates would appear rancorous to an outsider but to us it's normal Cypress blunt-speak.

At the end of the meeting, each VP has his or her allocation of dollars and headcount.

DC- What's the next step?

TJR- If the VP has a very large group they'll go through the same process of allocating headcount and dollars to their departments. However, as soon as the group is a reasonable size, the VP can meet with their managers and hammer out priorities by consensus.

Each group comes up with a list of what they want in priority order. The list shows what they are going to buy, given the budget, as well as what they are not going to buy. What is of interest to me is the cut-off point -- the line between the last few things they could buy with the budget and the next few things that missed the cut-off.

I can then check the validity of the first rough allocation of resources between groups. I look at the top three or four things Group A said they couldn’t afford and compare them to the top three of four things Group B said they couldn't afford. I can then make comparisons to things I know about. I'll see if the things Group A thought were pretty important but couldn't afford are comparable to the things another group had high on their lists. For example, if Group A is not going to hire people, I think they really should hire while Group B is doing stuff they can live without for a year then I know Group A is being under funded. I can then fine-tune the original budget allocations by shifting dollars from group to group.

DC- Can you give me an example of how this works?

TJR- Let me take you through an example where we needed to cut back on resources— it's exactly the same process. We recently needed to cut 3% of headcount. We asked every group to come in with a prioritized list of 5% of people they could let go. I could see people who just didn’t get saved and the people who just did get saved in each group. Even though we have 5,000 employees right now and I don't know every one of them, I know enough people to see if we are cutting at the right level in each group.

In one department, I recognized a few people on the list and it was clear to me that the cuts had gone too deep and it would hurt the company. I would rather lose money for another quarter than lose some people on that list. So I moved the line for that department from 3% to 1%. In another group I cut the whole 5% because it was clear that the last person on the list was very far from being critical for a company trying to make money.

DC- In closing, can you summarize the capital and headcount allocation process?

TJR- The capital allocation process has three parts. The generic allocation process downward, detailed allocation by the groups themselves, then adjustment of the lines to fine-tune the original allocation.

The groups come up with their list of priorities by consensus the same way the whole company did originally. I don't mess with those priorities in any one group, but I will move resources from one group to another.

This last step of fine-tuning is imprecise because your knowledge can’t be perfect. However, it can refine the judgment that set that first allocation which is the only mechanism we have.

DC- This is an interesting process and it makes a lot more sense than judging projects against some kind of ROI hurdle, which is what I learned in my MBA. What's particularly interesting is the third step when you fine-tune the allocations. Normally, we'd knock ourselves out trying to get the original allocations right when that's just not possible to do in a large firm. Thanks for sharing those ideas.