The competitive dynamics of the semiconductor industry have changed due to the increased adoption of portable devices. Intel (INTC +2.61%) was the leading semiconductor player before the rise of mobile and tablets. Although the company is still dominating the semiconductor industry, thanks to its PC chips, the industry is shifting toward mobile and as a result, the competition dynamics are changing. Foundries like Taiwan Semiconductor (TSM +2.05%) are becoming more relevant due to mobile chip manufacturing capabilities, followed by companies like Samsung and GlobalFoundries.
Fabless companies such as Qualcomm (QCOM 0.76%) are also emerging as semiconductor leaders, and Intel appears to have stepped away from the spotlight, for now. Mobile growth, processor architecture, and the business models of current semiconductor companies have brought about this change. Intel had the historical competitive edge and consistently maintained its lead in process technology, but that is also fading away amid Samsung and GlobalFoundries' 14nm partnership and Taiwan Semiconductor's 16nm and 10nm roadmap.
Current competitive scenario
Intel maintained its dominance in the semiconductor industry with its $48 billion semiconductor revenue and market share of around 20% in 2013, followed by Samsung. Among pure-play foundries, Taiwan Semiconductor maintained a lead with around 46% of the market.
Source: icinsights
The graph reveals that Intel had negative growth while Taiwan Semiconductor had the highest growth in 2013. The main reason for this was Taiwan Semiconductor's ARM manufacturing capability. Intel leads in process and node shrinking technologies, but ARM is preferred in mobile devices due to its low power requirements. The high expenditure associated with advanced process technologies makes it difficult for Intel to compete on a cost basis.
Moreover, ARM licenses its design, which saves R&D costs for the manufacturers. Intel must develop its own designs, which leads to high R&D costs. The company cannot benefit from licensing because the low-power devices industry has little or no interest in x86. All in all, the resistance to break away from x86 is the main reason for Intel's unproductive mobile operations.
Taiwan Semiconductor enjoys a leading position as a pure-play foundry due to its process technologies and manufacturing capacity. The company is set for mass production of 20nm, while GlobalFoundries and Samsung are stuck with 28nm for 2014. Intel did not take advantage of its 22nm when other manufacturers were stuck at 28nm because it did not offer ARM-based mobile solutions. The process node advantage materializes in mobile for ARM architecture, hence Taiwan Semiconductor holds an advantage over other ARM foundries in 2014.
Competitive scenario going forward
Samsung and GlobalFoundries announced a collaboration to increase the capacity footprint for the 14nm FinFET process technology. GlobalFoundries will license Samsung's 14nm FinFIT and the benefits are expected to be two-fold. First, 14nm will result in higher performance and power compared to high-node process technologies like 28nm. GlobalFoundries claims that by using 14nm FinFET, SoC performance gains can be increased by 20% and power requirements can be reduced by 35% as compared to 20nm planar process technology.
By Moore's Law, performance and power move in a favorable direction with an increase in the number of transistors per area. Therefore, 14nm FinFET will result in performance and power improvements. Secondly, production capacity will be increased, as the fabs of both companies will be utilized to implement the 14nm process. This can help in design wins from clients with supply concerns.
More importantly, this development is a significant threat to Taiwan Semiconductor and Intel. TSMC will be rolling out 16nm throughout 2015, and 10nm mass production is not expected until late 2016. This gives a huge competitive advantage to Samsung and GlobalFoundries. Samsung claims that the 14nm FinFET will be ready for mass production by the end of 2014.
This points toward flagship design wins in 2015 because of the marginal increase in performance of 14nm as compared to 16nm. 14nm is a slightly costly alternative, but performance is at the top of the pecking order for high-end and flagship devices. TSMC is expected to feel competitive pressure in the high-end arena. However, the lower cost of its 20nm and 16nm, as compared to the 14nm of Samsung and Intel, makes it an attractive option for medium to low-end devices in 2015.
|
16nm* |
14nm FinFET** | |
|---|---|---|
|
Performance compared to 28nm |
40% faster |
60% faster |
|
Power compared to 28nm |
55% less consumption |
60% less consumption |
*Taiwan semiconductor claims
**Global foundry claims
Using these claims as a reference point, it can be inferred that 14nm FinFET is faster and less power-hungry than 16nm process technology.
It is an uphill climb for Intel from here on. The only x86 advantage that it had in the mobile space has been lost because Samsung is ramping up for 14nm mass production. ARM 14nm will be preferred in the mobile space over x86 because, as seen in the past, 28nm ARM was preferred over 22nm x86. Now that the process advantage is lost, it seems unlikely that Intel will gain ground based on the x86 architecture or its current business model.
The problem does not inherently lie with x86, as a study by the University of Wisconsin researchers concluded that ARM and x86 are equally positioned to benefit from energy-constrained innovations in the future. The problem is that Intel took some time to reduce power requirements and ARM dominated the mobile market.
Now, the industry will not be willing to change unless Intel comes up with a differentiated alternative, not just an equal substitute, to ARM. Alternatively, Intel can embrace ARM and maintain its dominating position in the semiconductor industry amid its leading edge technology.
Bottom line
Taiwan Semiconductor appears to be dominating the mobile space this year. Going forward, Samsung and GlobalFoundries will be the biggest threat to TSMC due to leading-edge 14nm FinFET technology. Intel will not be able to capture the mobile market due to the shortfalls of x86. Samsung's foundry business and GlobalFoundries' revenue will grow in the high-end/flagship smartphone market. On the other hand, Taiwan Semiconductor's is likely to dominate the middle and low-end market.
