Earlier this week, Intel put out some solid earnings results. We do not need to rehash those here, instead want to focus on one particular aspect of their numbers – their long term gross margin guidance. In the latest quarter they reported 42.5% gross margins, but critically, they restated their goal for someday achieving gross margins in the 60%’s. Six months ago, most people including us, dismissed that goal as fanciful (at best), but now we all have to at least consider the possibility, slim as it may be, that they can actually reach those. In particular, we need to understand how Intel is re-organizing itself internally as it prepares to launch Intel Foundry Services (IFS) as a third party foundry for semiconductor designers.
Earlier this year, Intel announced that they would make some changes to how the do their internal accounting. Under the new structure, the product side of Intel and the fab operations will both have their own, separate P&L, and they will interact with each other on an arm’s length basis. Previously, the two sides blended their costs, with operations costs buried inside of product margins. This sounds like a trivial accounting change, after all the company will still report consolidated gross margins which will be unchanged, but this seemingly small change will have a big impact on internal incentives. Our core thesis on Intel is that their biggest challenge in the internal fight to change their culture, and so this ‘small’ change has big implications,
After earnings, we had the opportunity to speak with management about their results and we dug into the topic of how they build to that long term goal. What stood out for us in this conversation was management’s assertion that they are taking many ‘easy’ internal steps that will lead to gross margin improvements. Some of this is straightforward, for instance benchmarking fab operations costs against peers. Others are easy-sounding – making the product side pay for rush orders or ‘hot lots’. An obvious question is if these are all ‘low-hanging’ fruit why weren’t they doing these things already. The answer is that these changes are not going to be so easy.
Imagine you are an Intel salesperson, charged with covering one of the hyperscalers whose data centers are major consumers of Intel chips. The customer is about to make a big purchase decision – Intel vs. AMD CPUs – for what is likely a $1 billion order. If you win, you get the Cadillac and a giant bonus. AMD launched a new chip earlier in the year, but the Intel product is just coming out. The customer wants to build a 1,000 CPU system to test out both options under real work loads. Your product is in short supply, but you know if the customer uses the prior generation of Intel CPU AMD will win easily. So you go to the operations team and ask for a rush order of 1,000 parts. The operations people are reluctant to do this because the product is new, yields are low, so they will have to produce double the amount of wafers to get 1,000 good chips. Moreover, the customer wants a specific SKU or version of the chip, but the operations team is already set to produce a different SKU. To get those 1,000 parts will require them to shut down the fab for a few hours to re-tool, produce your chips, and then shut down again to re-tool for the original plan. When the fab costs $30 billion to build, a half day of downtime probably costs something like $10 million in deprecation. Under the old model, you did not care, because you know you can win that $1 billion order, and all those charges will become rounding error for the sale. Under the new model, you now have to bear the full brunt of those charges. Let’s say this order is taking place close to the end of the year, so you will have to take the charge on this year’s numbers, and will not get the purchase order until the next year. With your P&L in that shape, you will not get the steak knives, you will probably get fired, and someone else will get the benefit of that purchase order next year. But if you do not place the rush order you will likely lose the deal entirely. Reader, what would you do in this situation?
Of course this is highly simplified, but it does outline the serious nature of the cultural change that is about to take hold inside Intel. Put simply, Intel built its prior practices into its business model and sales motions. It gave the company a big advantage against their competitors. Absent this tool (crutch?) their sales team will have to rely much more heavily on product performance alone. This will affect their market share and revenue outlook in ways that are not entirely. As our hypothetical above shows, a lot will come down to individual choice, group dynamics and management culture. Will our hypothetical sales person’s boss be willing to take the long-term view? Will the operations team still be willing to provide any level of flexibility? How will senior management adjudicate disputes? Will management just agree to a one-time exception, which then becomes standard practice, erasing the whole point of the change? Accounting can be boring, but sometimes it leads to things that are very not boring.