07-04-2021 | | By Sam Brown
Now that Intel has decided to commit itself to the fabrication businesses the race is on between Intel and TSMC. What do node numbers mean, and what features are each company capable of?
In the past few years, Intel has faced multiple delays to new product lines as a result of node failures in its manufacturing division. Intel is quickly becoming the black sheep in the industry as more companies switch towards a fabless business model whereby they outsource their chip production while Intel continues to stick with manufacturing their own devices.
The exact cause behind Intel’s struggle against competitors such as AMD is a two-stage problem. The first stage that caused Intel grief was the fact that they were unable to produce sub 10nm devices reliably. However, the second stage, where Intel really felt the pain, was that Intel has tied its release of new hardware to the release of new semiconductor technology. With Intel struggling to produce the next generation of devices, Intel could not release their next-generation CPU technology either, and competitors took advantage of this.
However, Intel is finally working out its issues, did a bit of reorganisation, and has now slammed its credit card onto the poker table and doubling down on its bet. Despite some in the industry urging Intel to sell off its hardware production division, Intel has decided to not only stick with manufacturing their own devices but will open their tech up and become a manufacturer for other companies (the same model that TSMC follow).
When talking about semiconductor capabilities the term “node” is often used. In the past, node had a very specific meaning; the smallest feature size on the semiconductor device. For example, the 250nm node would allow for semiconductor devices with a minimum feature size of 250nm. Since this number often related to the number of transistors that could fit on the device, the smaller this number, the better a device was.
Fast forward to modern-day devices and the term node is no longer consistent or even helpful. While the node number may refer to feature size, it won’t always mean the smallest feature size or even relate to transistor density. For example, Intel 10nm allows for a transistor density of 100M transistors per mm2, while Samsung 8nm technology only allows for 64M transistors per mm2.
While there are other IC manufacturers such as Samsung, we will only consider Intel and TSMC as TSMC manufacturers AMD devices that directly compete with Intel, and Intel is offering services that will compete with TSMC.
So, the first comparison that should be made is who has the smallest technology? Answer – TSMC
Intel’s current technology lies around the 10nm mark with a transistor density of around 100M transistors per mm2. TSMC, however, is advertising devices in the 5nm range with a transistor density of 173M transistors per mm2.
The second comparison is which of the two is better equipped? Answer – TSMC
While Intel has far more experience than most companies on the planet producing semiconductor devices, experience from the 70s and 80s means absolutely nothing in modern manufacturing techniques. One piece of equipment, Extreme Ultraviolet lithography, is exclusively made by one company, ASML. According to ASML, TSMC will have anywhere between 30 to 63 such systems by the end of last year while Intel only has around 13 to 20. This allows TSMC to not only accelerate its production but also grants the opportunity to research on reducing transistor sizes.
The third comparison is who is the bigger company of the two? Answer – Tie
While Intel has a much higher yearly revenue than TSMC at $77.87 billion to $38.39 billion, it should be noted that Intel also designs and sells CPUs whereas TSMC solely focuses on manufacturing customer devices. As such, Intel may be at a disadvantage as TSMC has far more experience dealing with customers and manufacturing to their needs. Intel’s new business model will be competing in an industry it has never had experience in. What may give Intel an edge is the bigger name and more availability of funds.