There is a lot of discussion lately about whether Intel is an outlier when it comes to chip manufacturing capability. Most of the major chip makers these days have “outsourced” their manufacturing operations to companies who are producing leading edge devices for their customers. Does it make sense for Intel to do the same and go “fabless” to help it with its market challenges?
Just selling the fabs are not enough to solve all of Intel's challenges. They run deeper than that. The primary issue is that Intel lost its edge in innovation given the new market dynamics of AI powered sales and customer investments. Intel still has a dominant share of the PC space, and does OK in servers, but it really has a minimal share in the AI space where NVIDIA and AMD have superpowered GPUs that, more importantly, generate huge margins for them.
Intel's margins have been slipping for a couple of years and that's a major concern, as well as having a major impact on investment funding. And trying to catch up to Nvidia and AMD at the high end of AI will be tough. They are rightly concentrating on the mid end Inference Edge market, which will grow dramatically over the next couple of years. But let’s look specifically at whether or not it makes sense for Intel to get out of chip manufacturing.
Intel’s early dominance in the PC and data center space revolved around its advanced manufacturing capabilities as much as its design prowess. For many years, Intel was 2-3 years ahead in process technology. Through a lack of proper investment, and perhaps lack of focus, Intel let that slip away and became a follower rather than a leader. Of course, its late recognition of major market swings, especially AI, was a major contributor to its problems.
Many “fabless” rivals like Nvidia, AMD and Qualcomm seized on the new market dynamics and jumped ahead by using advanced chip production facilities like TSMC and Samsung. Indeed, AMD divested its fabs years ago with little long-term repercussions. So is “fabless” the way to go for Intel? It’s a little like a decision around public cloud vs internal data centers, or rent vs buy.
In the past few years, Intel has made a strategic investment in upgrading its manufacturing. Its “5 in 4” strategy (five semiconductor nodes produced in four years) aims to propel it to process leadership. I do think Intel can recover, but it will be two to three years of effort to get back to where they need to be in a leadership position, not just on the process technology, but also on the need to have mass scale from their own chips but also from their potential outside customers so the fabs are not revenue negative. Making Intel manufacturing a separate business unit requires it to not only supply in- house needs but have a range of external customers as well to keep the lines full.
Often unrecognized, its advances in chip packaging have also been a major thrust for Intel that has provided it with leadership packaging capability, especially now that chiplets and advanced SoCs are a critical requirement. But manufacturing operations are expensive and require scale and high volumes to be effective – something Intel has not been good at lately.
Selling the fabs gets them out from under a cash sink right now, but does not necessarily help them long term. And supply diversity is not a trivial thing, given the high concentration of chip making in the Far East and the potential volatility of that region.
Intel is highly affected by its recent financial issues which are leading to an impact on its investments. Rumors are that Intel will at a minimum scale back its investment of a new manufacturing plant in Germany. And it may scale back its build out of a new plant in Ohio and updates to some others in the US.
But since Intel has accepted a large sum of government funding (both federal grants and state tax rebates), it’s very likely the build outs will continue. Intel has been able to do some creative financing to help it weather the obstacles of the tens of billions needed to fund the projects, and likely will be able to continue to do so.
If Intel can truly up its game to a leadership position in the next 2 years, having its own manufacturing can be a major benefit to both its technology and ultimately its margins. The market seems to endorse a divestiture of the manufacturing operations, and it would offer Intel some needed capital and reduction in overhead and expenses in the short term.
RELATED: How to save Intel: No shortage of ideas surface
But it also puts Intel at a potential disadvantage as it’s dependent on using the same process technology as all the others, and therefore unable to leverage an advanced manufacturing capability that could be highly advantageous long term – provided Intel can truly regain its “ahead of the field” industry leadership.
No doubt this will be a difficult decision that management will need to make, trading off short-term financial gains for longer term technology advantage. But my take is that Intel has more to lose than to gain by letting go of its manufacturing.
Jack Gold is founder and principal analyst at J.Gold Associates, LLC. With more than 45 years of experience in the computer and electronics industries, and as an industry analyst for more than 25 years, he covers the many aspects of business and consumer computing and emerging technologies. Follow him on Twitter @jckgld or LinkedIn at https://www.linkedin.com/in/jckgld.