Intel Vs Samsung: The Foundry Scenario
Intel vs Samsung business is the talk of the hour. This article delves into how Intel’s Copy Exactly and Samsung will be affected by IDM2.0
Do weakening businesses copy the strategies of thriving businesses?
Is it a good idea to ape your biggest competitor?
Or is copying the strategy to go by when you have no more strings to pull?
Welcome back to the second part of the article, a continuation of “Why I think Intel is restarting its foundry business”. After publishing the article, I came across a few intense discussions with a few well-versed people in the industry. The outcome of their encouragement and discussions is the reason for this article.
We delve into-
- Intel’s “Copy Exactly”
- Samsung’s foundry model
- Samsung’s numbers and strategies
- Comparison with Intel’s IDM
- Insights and Conclusion
With Intel’s IDM there came a discussion about the biggest chaebol of Korea, Samsung. Samsung too was a vertically integrated Integrated Device Manufacturing unit before it created a subsidiary by the name of Samsung Foundries. This subsidiary belongs to Samsung semiconductors. Samsung like Intel was a vertically integrated company. A brilliant explanation about Samsung’s history can be found here.
What is the “Copy Exactly!” strategy?
Intel’s profound strategy to bring innovation to its table is called “Copy Exactly”. It emulates the different tiers of experimentation and research used in identifying a new technique. The different parameters required in each of these tiers are optimized to achieve the highest throughput. These may include even minor details like the color of the gloves, workbench height, and notepads used to write down the content.
All of these details are emulated exactly in the manufacturing sites without even a single aberration. According to this report, researchers spend at least four years at a technique and perfecting it. Employees from different sites all around the world, spend time learning this technique for at least a year, before meticulously transferring every small change to their sites. Intel’s “Copy exactly” strategy was first implemented in the Fab12 in Arizona, imported from the plant in Hillsboro, Oregon. Copy Exactly was implemented under Craig Barrett in the 1980s.
More information about the evolution of the Copy Exactly can be found here.
What are subsidiaries and how do they benefit a business?
Before we try to understand what Samsung’s business model for its foundry is, I just want to explain how subsidiaries help business.
As a refresher, vertically integrated companies are those which control all the aspects of the supply chain for a product. For example — a biscuit manufacturer procures raw material, makes biscuits, packs them, dislodges them to their flagship stores, and sells them there. They could also have retailers but are not solely dependent on them to reach end customers. The same analogy can be applied to chip manufacturing, where vertical integration is known as integrated development and manufacturing (IDM).
Now, the meter to adapt vertical integration to a company depends on the industry it belongs to. For the semiconductor/computer hardware industry, vertical integration has not been beneficial.
The two major issues faced were flexibility and loss of focus.
- When integrating all levels of the supply chain, a certain technology is adopted. However, when a disrupting technology comes to the fore, it becomes very difficult to adopt the same. Shifting to new technologies includes high costs, infrastructural changes, and human resource ramp-up.
In contrast, pure-play would have the resources to adopt new technologies and also diversify. They need not pipeline finance across the supply chain. The money, as it goes through the supply chain can diminish in value. In the pure-play there are no such hierarchies of movement, a single entry and exit allow for more money.
2. Secondly, the loss of focus in the work and management is inevitable in vertically integrated companies. The management team needs people who have expertise across the different steps in the staircase. Different human resource hiring ensues. There are even chances this may not click. This is what happened to semiconductor companies that were IDMs.
To overcome the disadvantage, they detached their foundry business as subsidiaries.
What is Samsung’s foundry model?
Samsung was an IDM till the year 2017. It explicitly spun off to avoid “conflict of interests” with its customers and diversify its foundry portfolio.
The key benefits of having a subsidiary are diversification, independence, and finances. Subsidiaries can open up shares in the markets to raise funds. Consequently, (obviously) it helps to drive funds in the single-tier instead of rotating it across the different echelons. Having a subsidiary will diversify the customer portfolio, removing the “conflicts of interest”. A few customers can feel that in the case of an IDM, the customers would have had a few qualms about collaborating with them. The company can completely unleash its potential to serve customers. This news article from 2018 announces Samsung foundry’s entry into the market.
Samsung ticked all the above-mentioned and more. It posted a profit of 3.4 billion USD in the year 2019, up 19.3% from 2018. Samsung’s market share is about 17%, according to the same source. The main customers for Samsung are Qualcomm and Nvidia.
In 2016, Joong Shik Yoon, Executive Vice President of the Samsung Foundry business talked about the business’s engagements and roadmap. He mentioned collaborations with EDA accelerators, self-driving cars, and IoT devices. For instance, Samsung and Synopsys have collaborated to bring out the fastest 3nm process sign-off and delivery in 2020. They have also partnered with Tesla to deliver 5nm chips for its autonomous products. Ansys and Samsung have collaborated again to accelerate FinFET nodes.
Cut to 2021. Pat Gelsinger delivers his keynote on IDM 2.0- Intel’s new strategy to prop up its flailing business. A comprehensive explanation of Intel’s IDM2.0 can be found here.
Comparing this strategy with Samsung’s, it only feels Intel has done what it is best at doing –“Copy Exactly”. It has emulated Samsung’s ideas into its new foundry subsidiary. The only difference being Samsung realized it early that spinning off will prove more fruitful though, it did not suffer any losses. However, Intel had a hard time before it could make this move.
Here are some of the things Intel plans to do-
1. Intel plans to manufacture ARM’s RISC-based CPUs similar to how Samsung has committed to Qualcomm and Nvidia (that use RISC).
2. Intel plans to collaborate with EDA providers to accelerate the process methodologies leading to quicker delivery to customers. This allows for standardization in the industry.
3. Intel also plans to become the leading manufacturer of chips in the US and the European Union. This gives a déjà vu to how Samsung became the leading manufacturer in Korea and slowly expanded to the world in consumer electronics. Moreover, it has also disrupted TSMC (the market leader in foundry business) to some extent in such a short stint. Pretty commendable, right?
4. Samsung, in early 2021, announced a potential fab in Austin, Texas. Intel has followed on with a similar announcement in its IDM2.0, having two new fabs in Arizona. Though the factor can also be attributed to the Chinese mistrust issues, it does not seem very unlikely that one company is indeed following the other’s footsteps.
5. Intel has let out a recent statement that it will start making chips for automotive within the next 6–8 months. Ok, need I say more?
The potential growth of foundry business is posted at 12% YoY with total revenue of 92 billion USD. Samsung’s YoY is projected as 20%. Intel is a new player in the town. Though his presence will not affect TSMC, it will be a hurdle to Samsung. If Samsung can employ its strategy of “read market, fail faster, and rise quicker” Intel will not become a threat to them.
Here are my two cents-
1. Intel and Samsung can collaborate on automotive chip production to accelerate their gains. It can be a symbiotic relationship where Intel can design the automotive chips. Samsung can manufacture them. Or, signing the deals, they can partner to give double the throughput in the manufacturing only without spending in infrastructures.
2. It can be coincidental that Intel “Copied Exactly” Samsung’s strategy. The devil could lie in the details which Pat Gelsinger did not unveil. Differentiation between the two companies is well-hidden(at least for now until Intel completes the spin-off).
3. For Samsung not to be engulfed by Intel, it needs to get its highest share of customers in consumer communication electronics, and high-performance computing hardware, that promise high returns.
If you have more thoughts, leave them in the comments below.
