Intel Finally Solving the Right Problem

When I first heard that Intel was replacing their current CEO, Bob Swan, my first thought was “rearranging the deck chairs on the Titanic”. After all, you don’t get into a situation like Intel’s without some persistently wrong management choices. The board, despite being filled with some proven intelligent people, doesn’t understand Intel’s technical details, nor has it valued the depth of knowledge embodied in long-time employees. It does, however, desire a profitable company, and this is where Pat Gelsinger arrives on the scene.

The Wrong Problem

You can’t fix a company until you admit you have a problem. Everyone understands this, even the Intel board members. Intel’s woes have been in the news for years at this point, and the company has responded at each setback. Oftentimes, those responses have been drastic. After all, the entire company knew it had a problem. So, why didn’t it just fix it? It was fixing the wrong problem.

To understand, we should look at the expertise of the Intel leadership. Intel’s board has a history of picking very smart and talented people even though many would disagree. It’s easy to armchair quarterback from the sidelines. After all, we have the benefit of hindsight as well as a completely different perspective.

Each Intel CEO was chosen specifically to solve problems. Starting with Paul Otellini, Intel picked a non-technical CEO because it was solving a non-technical problem: marketing. In hindsight, we can see that Otellini was the start of Intel’s current woes: R&D failure. Remember, however, that Otellini was picked to deal with the end of Moore’s so-called Law. Steady advances in chip frequency had hit a brick wall (remember the Pentium 4?) and Intel needed to win with something other than the easy gains from die shrinkage. Remember when Apple chose Intel CPU’s? That has been largely been attributed to Otellini’s business expertise. Otellini wasn’t stupid at all, and he did bring much needed business expertise to Intel, but he did start the trend of Intel fixing the wrong problem.

Brian Krzanich appears to have been chosen to fix the right problem: chip manufacturing. After all, his education was as a chemical engineer and he previously worked at Intel as a fab manager for the Chandler, Arizona plant. However, his career rise in Intel was in the logistics supply chain, not in a technical role. He did appear to recognize the need to target the mobile market but failed to create the teams that would make it happen. Instead, he laid off thousands of valuable employees when his strategies failed and ended up losing irreplaceable process knowledge. His tenure ended with a scandal involving an extramarital affair with a subordinate, however, his real failure was to ruin the 14nm and 10nm processes. Cheap contractors may be “agile”, but Intel repeatedly encouraged its most valuable assets (employees) to retire or move to other companies. Near the end, he did appear to make some high-profile technology hires (e.g., Jim Keller and Raja Koduri) but it was simply too late.

This brings us to Bob Swan. Swan is definitely an expert in business. He has an MBA and an extensive background in finance roles from eBay, General Atlantic and then Intel. So, how did he get the CEO role? It was probably because it was an emergency replacement that just didn’t get replaced as quickly as Intel needed. Bob Swan is a steady and trustworthy guy, which you want in finance, but what did he do with his most valuable resources? Apparently, he was unable to resolve fundamental political problems (e.g. in-fighting between Keller, Venkata “Murthy” Renduchintala, and Raja Koduri) and failed to see through the foolish promises about 10nm from Murthy or GPU development from Raja. Honestly, he explicitly stated that he didn’t want the job in the first place, and that was likely a good assessment. His skill set wasn’t what Intel needed. He wasn’t technical, nor was he politically astute enough to manage the technical subordinates that could have solved the issues.

The Right Problem

Thankfully, we come to Pat Gelsinger.  Intel’s  board has to have seen the successes of Lisa Su at AMD. She is not just a good business mind; she is also a first-rate engineer. So, the board has seen a good model of what they need.

Gelsinger has a strong technical background as well as a deep understanding of what made Intel function. It isn’t simply logic design, or chip fabrication, or marketing, or business deals, or any of several facets that were handled by recent Intel CEO’s. Pat Gelsinger does have technical background within Intel, but he has a matching business background gained as CEO of VMWare.

However, Gelsinger grew up as a farmer. Yes, a farmer, in an Amish part of Pennsylvania. His first Intel job was as a quality-control technician at age 18, where he simultaneously earned his degree in Electrical Engineering from Santa Clara University. He immediately worked on his master’s degree from Stanford University (in Electrical Engineering and Computer Science) while simultaneously working on the 80386. Not stopping there, he then becoming the chief architect of the 80486.

That’s a boot-strap career rocket, boys and girls, as Gelsinger had to be outstanding at each of those stages. He wasn’t simply a “good” engineer. Going from Santa Clara University to Stanford University is no small feat, as is becoming a chief chip architect after a short stint as logic designer on the 80386. As Tim Sweeny noted on Twitter, Gelsinger co-wrote the book on programming the 80386 with the chief chip architect. There is a reason Gelsinger became Intel’s CTO in 2000. Just by coincidence, Intel was really good at chip manufacturing during that time period, and he mysteriously moved on when Intel chose its first non-technical CEO.

I’m not saying that Gelsinger was the reason Intel was so good during 2000-2005. Things aren’t so cut and dried. After all, Intel is a huge company and owes its successes and failures to that huge number of employees. However, culture matters.

Gelsinger doesn’t appear to have a background in semiconductor fabrication. Yet, the people don’t understand how the sausage gets made in Intel. Logic designers work very closely with the fabrication engineers. This was especially true during the early days such as the 80386 and 80486. Logic engineers had to intimately understand the transistor and interconnect parameters. So, they were exposed to the successes and challenges from the fab group. In reality, the fab group really did and still does drive the logic group. You can’t design something without knowing your materials. In addition, remember Gelsinger’s first job?  Yes, it was quality control technician. You can bet he remembers how important it is to make things right.

The Future

Unfortunately, this is going to be a huge test for Gelsinger no matter how smart he is. The good news is that Intel has acquired an excellent candidate with a skill set that matches the real problem: reliable excellence. There are plenty of smart people within Intel, but it appears that Gelsinger will be the leader that focuses, enables, and unleashes that potential.

He was a farmer, so he knows how to make things grow and certainly remembers the “little guy” in the trenches. He is brilliant and remembers the importance of developing and keeping your employees. He certainly remembers countless instances of cutting through the smoke when dealing with overconfident managers. It doesn’t matter how good a logic design is if you can’t make the required wafers every time. That requires some excellent quality.

I’m not a fanboi of any of the tech companies, but I’m rooting for an Intel win here. Everyone says it because it’s true: we need good competition.

So, good luck, Intel.