TSMC’s 7nm node will soon be removed by the 5nm node, but it is the most widely used in the industry, as TSMC is the factory used by the vast majority of “ factory-less ” semiconductor companies. or non-manufacturing. As is the case with AMD, currently TSMC’s largest customer for its 7nm node.
The first step is the wafers
The first thing needed to be able to make a processor is wafer making where the different processors are printed together before being laser cut and extracted from the wafers themselves. Which are made from silicon ingots, which are similar in shape to what you can see in the image above.
From each silicon ingot is taken a certain number of wafers, which correspond to the minimum order per processor. In this state, each 300mm diameter slice costs an average of $ 400, but there are a series of additional costs to be added. To this cost must be added other additional costs such as taxes, the operating cost of the factory, the wages of the workers and especially the development of new manufacturing nodes, which are more and more expensive to develop. The final cost of platelets is therefore much higher than this initial amount of $ 400 and can go up to several thousand dollars.
The cost of the TSMC 7nm node
TSMC never disclosed the costs per wafer, so the source in this case is Sophie Wilson, an engineer at Acorn Computers, who in a presentation by her disclosed the costs per wafer manufactured under TSMC’s 7nm node. This presentation was featured on Ian Cutress’ The Tech Potato channel.
If we look at the table, we will see that the cost per tranche is $ 5,859.28, which corresponds to a cost of $ 1.52 for 100 million logic gates. To this must be added two problems concerning the costs of each particular processor.
Problems That Increase Manufacturing Costs
The first problem with new crafting nodes is the so-called Dark silicon, a phenomenon in which the area where the processor is to be idle grows with each new generation, it means that when implementing a new design or transporting an old one, not only transistors must be taken counted as active, but also inactive. As a result, the different models no longer fit as before and require larger areas than expected.
The second is the defect rate, which at TSMC’s 7nm node is 0.09 per cm2, so the bigger a chip, the less good units will come out. For example, a Navi 21 GPU, being a much larger chip, will not only mean fewer chips per wafer, but the number of flawless chips that come out will be in less quantity and therefore be much more expensive.
In the specific case of AMD, it gets more profitability from the CCDs of its Zen 2 and Zen 3 than from manufacturing AMD Radeon GPUs, and it even gets more profitability from an SoC than from its GPUs due to the size of each. So all of this influences the choice of chips to manufacture at all times.
Why does TSMC not officially disclose the price of its pads?
Depending on when a chip is manufactured, one release date or another can be scheduled. You have to take into account that there are dates of the year when consumption is skyrocketing and it is important to have your product ready for those dates, which is why companies are fighting for their bullets are printed on TSMC brochures at certain times of the year.
Rumors have it that although TSMC has an average price to produce in its factories, what it is doing is sort of an auction and gives benefits to products that perform more effectively and for which its marketers will earn more money. ‘silver.
In the case of AMD, this is curious, because if we look at its most successful product in recent months, it is not a SoC for PC, but one for PlayStation 5, but the margins for AMD are very weak. Instead, Apple has huge margins, allowing them to pay more money per installment, which helps TSMC consolidate prices in the long run. Therefore, not all markets make the leap to a new manufacturing node, but those with the highest profitability per chip.