Intel is undergoing a significant strategic shift from being predominantly a product company with manufacturing to have two distinct strategic business areas. The company is untangling the manufacturing part from the product part by creating Intel Products and Intel Foundry Services (IFS).
From a strategic perspective, this is an ambitious move that is new to Intel. The company, used to being the 500 pound gorilla in the semiconductor market, has been used to being far ahead and dictating the rules of the industry. Both from a product and a process technology perspective.
The strategic move might be ambitious but it is also driven out of necessity. Intel’s financial performance has been subpar lately, not something the company or its stakeholders are used to.
In just two years, Intel’s operating profit has almost been eradicated, which is more than sufficient reason to make a strategic change.
What is Strategy?
If you ask a room full of business people to define strategy, they will all be able to do it. Unfortunately they will not all have the same definitions as strategy is complex concept with many possible definitions that all can be right at the same time.
The definiton used here is Strategy in its simplest form. It is about the future, action & winning in a market setting. It boils down to answering 3 basic questions:
What is going on?
What can we do about it?
What are we going to do about it?
With this simple approach, the background for Intel’s strategic shift is going to be analysed
What is going on?
The subpar financial performance that undoubtedly is the reason for Intel’s strategic change is the result of a number of underlying issues, that needs to be addressed in order to develop a sound strategy.
The first thing needed is an honest assessment of the strategic position and what is changing in and around the company. It was refreshing to hear Pat Gelsinger being very frank when presenting the split of Intel Products and Intel Foundry Services (IFS).
He highlighted some strategic area’s where Intel had made bets that turned out to be suboptimal. It is easy to call them wrong, but that is too simplistic. They are choices made on the information available on the time, that turned out to be counterproductive. We have added a few of our own.
Underinvestments compared to leader
Having been a product company, it is not surprising that the product business units have been dominating and manufacturing takes the back seat.
Product development is predominantly driven by R&D spend. Manufacturing and process development requires both R&D spend and significant Capital Expenditure in equipment and new factories. Intel IFS will need to measure itself against its main competitor, TSMC.
Intel and TSMC’s Capex can be seen below. As 90% of Intel’s Capex spend is allocated to manufacturing, the two are comparable, within a brick or two.
Having outspent TSMC, Intel has fallen significantly behind on Capex over the last 4 years.
Pat Gelsinger admitted that Intel had been underinvesting for a period of time. It is not unusual for corporations to “borrow” from the future when financial performance is down in order to satisfy shareholder returns, but it always comes with a cost later.
EUV was the winner
From a technology perspective, Pat Gelsigner admitted the loss of technology leadership that Intel had enjoyed earlier. He was clear that Intel was late to adapt the Extreme UV technology and that it had cost the company its leadership.
For IDM’s and Foundries there is a direct relationship between the technology competitiveness and the profitability as can be seen from the chart.
Intel’s own view is that they will be back in the leadership position with their new Intel 18A process due to launch within the near future. This is a prediction by Intel and while it might be sound, it is not the same as commercial success. That requires more than technology.
While there is no reason to blame Intel for choosing another path than EUV, there were signs in the market data that EUV was on the rise.
In a couple of sprints, ASML was starting to take significant marketshare in the Semiconductor tools market and as the company was the undisputed champion in EUV, this was a signal. ASML now has a third of the market of the top 5 tool makers as a result of the move towards EUV by TSMC and the memory IDM’s.
It could be that Intel saw it but had decided that their chosen path was better.
Mouse and Elephant product strategy.
In 2018 we did a strategic review of Intel that at that time had a very clear product based strategy with well defined applications and markets.
Over time, the business would expand into new emerging products and markets.
While the strategic areas were well chosen and all become high growth, Intel did not manage to grow each of the new business units to an appetising level. The problem is always the same, the new business unit looks small when compared to the Client Computing Group and to the Data center group.
Intel is not well known for patience with new business areas and has made reorganisations quite often. The last major one was two years ago, while the patience with the Advanced Graphics Group only lasted a year.
It looks like Intel has given up the struggle to find the third strategic leg in a product strategy and is pursuing it in a manufacturing strategy instead.
Divisional Bullies
With an overwhelming share of the business, CCG and DCAI are calling the shots inside Intel. Anybody that has been in corporate semiconductors knows that resources are allocated according to a business units past rather than its future.
A divisional structure gives great benefits to senior management. The divisions will fight with each other over resources and drive the corporation forwards. It is easy to manage as the divisions keep watch over each other and you can HireFirePromote as a result of business results.
The drawback is that the large divisions, bully (no critique, just the way a divisional structure works) their way to get what they believe is their fair share of resources and attention.
Intel indirectly admits this by highlighting manufacturing cost savings of a market based strategy:
It looks like the product divisions really got their way with the poor manufacturing division. A market based model could save up to 5B$ a year.
Intels strategic choice
The second strategic question related to what options Intel has within its reach (based on the companies capabilities). The decision of breaking the company up into a product based company and a manufacturing company is well within Intels capabilities and an obvious choice from a strategic level.
There might have been other options open to Intel, but this is the answer to the third strategic question: To split Intel into two companies, Intel Products and Intel Foundry Services, to address the changing environment.
Obvious does not mean easy. This is a very brave (Intel has always been a product company) and somewhat risky (still need to gain technology leadership) endeavour.
The strategy chosen should address all of the “What is going on?” issues we have covered above and adapt to the new reality. Pat Gelsinger and team have been quite honest about their assessment of the current situation and Intels shortcomings, and have set sails to catch the winds of the new reality.
The strategic objectives
As part of the overall strategy, Intel highlighted a number of strategic objectives that need to be achieved:
Shine a bright light on cost opportunities
Attract smart capital
Bring the wafers home
5 nodes in 4 years
De-risk process for external customers
The release of IFS data this week, shows that Intel is serious on being transparent with the division of IFS and Intel Products. It also helps prevent divisional bullying as the perfomance of both companies will be very visible and it will be impossible to curry favour with the large product divisions.
The transparency also highlight the quite difficult starting position Intel has for this strategy.
At the presentation, Intel highlighted that it already had 5 customers for its new 18A process which likely means that Intel has a path to meaningful revenue soon.
As a result of the IFS split, Intel is now entering a new market. For most of the history of the Semiconductor market, Intel has been the 500 pound Gorilla dominating the market briefly shadowed by Samsung when memory pricing was high. But now Intel is entering the foundry jungle where TSMC is the gorilla and Intel has to start out as the monkey. This will be a new role for Intel and will require new skills and attitudes.
Pat Gelsinger and team has sold this strategy incredibly well and even got the Biden administration to chip in with somewhere between 8-30B$ over the next years via the chips act. Gelsinger calls this “Smart Capital”, while Dire Straits calls it “Money for Nothing”. Nevertheless, it is a good choice for all parties involved and Intel has already achieved this important goal.
As shown earlier, Intel had a capex spend of just under 26B$ compared to TSMC’s CapEx spend of just over 30B$. With 90% of its CapEx allocated to Manufacturing, Intel is 7B$/year behind TSMC as the moment so the “Smart Capital” will close the gap to TSMC for the next few years.
“Bring the wafers home” highligt the likely humiliating fact that Intel is using TSMC in order to manufacture competitive products. There are limited facts about the scale of this business as it is not exactly covering Intel in glory.
With the new transparency in the IFS numbers, it is now possible to get an idea of what the scale of this business is.
A $ IFS revenue should roughly equate to a $ Intel COGS and as the comparison shows, there is quite a lot of COGS that is not IFS revenue. The OSAT cost for 2023 is 953M$ so that leaves a lot of COGS uncovered. This is likely TSMC wafers Pat would like to get back into IFS.
The starting point of Intels strategy
The starting point for Intel strategy looks like this:
Intel is under pressure on its home turf. The data center business is shrinking causing the significant drop in operating profits. While Intel will have meaningful revenue in foundry soon, it is not additive to its business. Intel will have to succeed in the AI GPU market.
We will continue to tell the story about Intel’s strategy with more data and charts at a later stage when more information is available