JasonDoiy
Investment Thesis
It is clear Pat Gelsinger is driving – reinstating – a culture of excellence at Intel (NASDAQ:INTC) across the organization. He just introduced the second phase of the IDM 2.0 model: the creation of an internal foundry model.
While interesting, it is not sure what problems this will really solve. In the last few months, Intel has given a masterclass in financial engineering, and this just seems like the next step. If the real engineering is at the same level, then surely Intel will again become the top dog of the semiconductor industry.
Background
A while ago I detailed Intel’s “world-class” financial engineering, as it introduced a novel model for co-financing its expensive semiconductor fabs. Intel will basically get $15B from Brookfield to build a $30B fab (actually two fabs worth $15B), but the catch is that Intel will only own a 51% stake (read: Intel will only receive 51% of the profits from those fabs). Ergo: there is no free lunch. Hence I did not see any investor benefits for this model.
Internal foundry model
The latest Intel press release bombastically states that “Intel embraces an internal foundry model”. However, after reading it, it is actually not really clear what this will really entail, or rather what benefit this would have. That is why I am categorizing this as Intel’s latest stint on the financial engineering side – I wouldn’t be surprised if it was actually the CFO that came up with this. But let’s review what Pat Gelsinger said anyway.
Pat Gelsinger, in what is quite the plot twist, revealed that the IDM 2.0 (integrated device manufacturer) strategy that he introduced last year (see Intel: The Empire Strikes Back) actually contained two distinct phases. The first phase was what was announced back then: the 5 nodes in 4 years (ignoring the first one, it is actually 4 nodes in 2 years). Pat Gelsinger reiterated Intel remains on track to this goal.
The second phase starts “today”, and it is the internal foundry model. This basically means that Intel (its own product lines) will now become an internal customer (internal customers) of its own fabs – it will be just one of the many customers that Intel Foundry Services will begin serving over the next few years. So foundry is now the new IDM.
This is a significant evolution in how we think and operate as a company, but the systems and infrastructure that served us well in the IDM 1.0 world will not enable us to achieve the full potential of IDM 2.0.
The explainer that then follows does not really explain that much, but the usage of “guardrails” that will now exist between the different teams lead me to suggest above that it was perhaps a Dave Zinsner idea. The words that then follow could have equally come from any marketing checklist: address structural inefficiencies, driving accountability and make decisions in real-time. In any case, the point remains that management is now relegating its own design teams to equal footing to its foundry customers (which is actually quite brave and yet again shows the commitment that Intel has made to its foundry business).
To that end, Intel says it will create a foundry accounting model that will bring transparency in things like gross margins, and allows Intel to benchmark itself against other foundries. To do so, it has established an “IDM 2.0 Acceleration Office”, led by the new Chief Business Transformation Officer.
As an example, design teams would be able to consider the impact of running an additional product stepping on their gross margin, while the fab team would be able to assess its factory output. The goal ultimately is to maximize output, reduce costs and shorten design cycles.
Implications
Ultimately, the internal foundry model does seem a bit like an exercise in futility. What matters at the end of the day is if the process technology roadmap remains on schedule (which it is), if it can attain high yield (remains to be seen) and whether the products that actually drive Intel’s revenue will use this technology from day 1 when it is ready (no Sapphire Rapids situations).
Nevertheless, the latest introduction does fit in the theme since Pat Gelsinger became CEO: to restore Intel’s culture of technical excellence. To realize this has required adopting new tools, practices and models – besides just adding sheer new R&D dollars and employees.
Alongside initiatives like the next evolution of our tick-tock model and our Gladius program for driving excellence in product development, embracing an internal foundry model will enable us to deliver the competitive cost structure and predictable cadence of leadership products essential to our success.
Investor Takeaway
During the heydays of Tick-tock, Intel’s process and architecture development teams worked together in unison to establish Intel as the firm driver of Moore’s Law, for well over decade well ahead of any and all competition. So clearly, there is some historical evidence that IDM does work.
To that end, the new “internal foundry model” where Intel will see itself as a foundry customer of its own fabs could perhaps be seen as gimmick. It was just a bit over a month ago with that Brookfield that I thought that was the closest Intel would ever get to spinning off its own fabs, but with this new move Intel has proven me wrong.
In fact, some people may start to speculate that once Intel has for all intents and purposes separated its design and manufacturing teams between this “internal foundry model” wall, nothing would prevent Intel anymore from actually spinning off its fabs. Or its design teams.
Nevertheless, for now this will not change the underlying economics of the company, which as I recently detailed remain unchanged and is the cornerstone of my bullish thesis.
In summary, the success of a semiconductor company can be measured by the transistor density of its process technology (Moore’s Law) in combination with the yield achieved on said technology. Since Intel remains (per Pat Gelsinger) on track to regain industry leadership by 2025, it remains in pole position to become top dog. How it internally cooks up the products (“internal foundry model”) that come to market based on this technology is utterly relevant, as long as the products actually come to market.
Be the first to comment