Abstract:
In the process of obtaining Qualcomm's 2nm chip foundry orders, Samsung Electronics seemed to have finally crossed the technical threshold that had long plagued the cooperation between the two parties. However, just as the technical problems were gradually being resolved, Samsung and Qualcomm had disagreements over chip foundry prices, resulting in the two parties being unable to formally reach an agreement.
For some time in the past, Qualcomm has not handed over its new generation flagship chips to Samsung. One of the core reasons is that the yield and stability of Samsung's advanced process have not met Qualcomm's requirements. Therefore, Qualcomm continues to hand over the 2nm chips of Snapdragon 8 Elite Gen 6 and Snapdragon 8 Elite Gen 6 Pro to TSMC and use TSMC’s N2P process.
If the latest news is true, Samsung has made significant progress in the 2nm SF2 and second-generation 2nm SF2P processes, and the technical obstacles that previously affected Qualcomm's cooperation have basically been resolved. Samsung’s own Exynos 2700 is also regarded as an important proof of the maturity of its 2nm process, and the performance of this chip has received relatively positive reviews.
Previously, the reason why Qualcomm has been slow to adopt Samsung's 2nm foundry solution is mainly because Samsung's 2nm process yield has not been able to reach the level required by large customers for large-scale commercial production. For chip design companies like Qualcomm, it is not enough to just have the nominal node of an advanced process. Wafer yield, performance stability and large-scale mass production capabilities also determine the final cost and supply reliability.
Samsung has been working hard to improve the yield rate of the 2nm process in recent years and is trying to further improve performance and production efficiency through the second-generation 2nm process. As these technical issues are gradually resolved, Samsung was expected to become an important foundry partner of Qualcomm again.
However, after the technical issues were resolved, new differences arose between the two parties on price.
According to reports, Qualcomm wants Samsung to further reduce the foundry price of 2nm chips, but Samsung is unwilling to accept this request. Samsung believes that the current number of 2nm chips that Qualcomm plans to produce is not enough to support a significant price reduction, so there is no need to sacrifice too much profit in order to obtain this order.
This also reflects that Samsung’s foundry business strategy is changing. In the past few years, in order to expand the scale of wafer foundry customers, Samsung has adopted a more aggressive pricing strategy and used low prices to win customer orders. However, as the demand for artificial intelligence chips has skyrocketed in recent years, advanced process production capacity has become increasingly tight, and Samsung has begun to gradually increase its bargaining power.
In August this year, Samsung was revealed to have increased foundry prices for some advanced process wafers, with new orders increasing by up to 15%. This means that Samsung is no longer as eager to obtain orders through low prices as it was in the past, but has begun to pay more attention to the profit margin of the foundry business.
Samsung has also recently received large orders from important customers such as Tesla and Broadcom, which further enhances its confidence in the foundry market. With other large customers, Samsung does not need to make significant price concessions in order to win Qualcomm's relatively limited 2nm orders.
For Qualcomm, the situation is different. Qualcomm is currently facing the problem of rising chip business costs, and TSMC’s 2nm process itself is expensive. If Samsung can be involved in foundry, it can reduce production costs through a dual-supplier strategy and reduce dependence on one TSMC foundry.
Therefore, it is not difficult to understand that Qualcomm hopes that Samsung can provide more competitive prices.
The manufacturing cost of advanced processes is inherently very high. 2nm wafers require the use of extremely expensive advanced manufacturing equipment, including ASML's high numerical aperture extreme ultraviolet lithography equipment. Fabs must balance huge equipment investments, R&D costs and higher production risks, while lower wafer yields will further push up the actual manufacturing cost of each chip.
If Samsung's current 2nm process has reached a sufficiently mature level of commercialization, then its necessity to continue to win orders from Qualcomm at low prices will naturally be reduced. In turn, if Qualcomm believes that Samsung's 2nm production capacity can become an alternative to TSMC, it hopes to use competition between the two suppliers to drive down manufacturing costs.
The current negotiations between the two parties have reached a deadlock.
What’s even more troublesome is that even if Samsung and Qualcomm reach a price agreement immediately, it will still take some time to actually form a legally binding foundry contract. Qualcomm has already placed an initial order with TSMC, and the production of flagship chips must strictly match the product release rhythm of mobile phone manufacturers. Therefore, even if Samsung finally gets the order, it may be difficult to change the overall production arrangement of this year's flagship chips in the short term.
This means that Samsung is more likely to become Qualcomm's backup supplier in the future, rather than immediately replacing TSMC as the main production base.
Qualcomm’s potential production scale of its current 2nm chips may also be affected by market demand. As the global memory market is experiencing supply constraints, DRAM and NAND flash memory prices continue to rise, and smartphone manufacturers are facing significantly increased overall cost pressure. As a result, some mobile phone manufacturers may reduce the scale of adopting new generation flagship chips to control the cost of the entire machine.
If Qualcomm expects shipments of Snapdragon 8 Elite Gen 6 and Snapdragon 8 Elite Gen 6 Pro to be lower than initially expected, then it will naturally not be willing to pay exorbitant prices for 2nm foundry. After the number of chip purchases decreases, the space for wafer factories to dilute fixed costs through scale effects will also decrease, which further intensifies the differences between the two parties over the price of a single wafer.
Previously, there was news that Qualcomm may adopt a parallel strategy of new and old process chips and continue to retain the 3nm product line in addition to the new generation of 2nm Snapdragon chips. This can not only meet the needs of some flagship mobile phone manufacturers, but also reduce the cost pressure caused by the full shift to 2nm.
If this strategy is finally implemented, Qualcomm's demand for Samsung's 2nm production capacity may further decline, which also explains why Samsung believes that the current order size is not enough to support a significant price reduction.
For Samsung, another important change is that its foundry business is gradually getting rid of the past "price-for-order" strategy. With the addition of large customers such as Tesla and Broadcom, and the continued increase in demand for artificial intelligence chip manufacturing, Samsung has gained more bargaining space. Rather than taking a limited-scale order at a lower price, Samsung may prefer to reserve its limited advanced process capacity for customers with higher profit margins.
However, this information still mainly comes from industry sources. Samsung and Qualcomm have not yet officially announced the final results of the 2nm foundry agreement between the two parties. Therefore, it is still not fully confirmed whether Samsung has completely solved the 2nm yield issue that Qualcomm has previously focused on, and whether the two parties can finally reach an agreement on price.
If Samsung finally succeeds in winning Qualcomm's 2nm order, this will be an important breakthrough for its advanced process business. Qualcomm is one of the world's largest mobile SoC design companies. Its flagship chip orders are not only considerable in scale, but also have a strong industry demonstration effect. Once Qualcomm confirms that Samsung's 2nm process is mature enough, other chip design companies may also re-evaluate the possibility of Samsung as a second supplier.
On the other hand, if the two parties are ultimately unable to cooperate due to price issues, then even if Samsung has crossed the threshold previously set by Qualcomm at the technical level, it will still be unable to truly translate its technical advantages into actual orders from Qualcomm.
Therefore, the key to this negotiation has gradually changed from "whether Samsung can do 2nm well" to "at what price both parties are willing to cooperate." Samsung is trying to prove that its 2nm process is mature enough, while Qualcomm hopes to use competition between Samsung and TSMC to reduce manufacturing costs. Technical obstacles are disappearing, but the game of commercial interests has just begun.
Comments