The FOPLP technology is derived from TSMC's InFO (Integrated Fan-Out Package) technology invented in 2016, and further derived from the FOWLP0 (Fan-Out Wafer-Level Packaging) technology. FOPLP technology significantly improves the number of chips produced and the cost advantage by replacing the IC substrate from the traditional 8-inch/12-inch wafer carrier. This technology takes RDL (rerouting layer) as the core, which replaces the role of the substrate transmission signal under the traditional package, making the package design more flexible and efficient.
With the rapid development of artificial intelligence (AI) technology, the performance and cost-effectiveness of graphics processing units (GPUs), as the core components of AI computing, have become the focus of industry attention. In this context, advanced packaging technology is particularly important. Among them, fan-out panel level packaging (FOPLP) technology is gradually becoming the new favorite in the field of AI GPU packaging due to its unique advantages.
With its advantages of low unit cost, large package size, and high transmission efficiency, FOPLP technology is an ideal choice to solve the packaging challenges of AI GPUs. By adopting FOPLP technology, AI GPUs can achieve more efficient signal transmission and more flexible package design, thereby meeting the needs of AI applications for high performance, low power consumption, and low cost. In addition, FOPLP technology also helps to improve the thermal performance of AI GPUs and extend the life of the chip.
Figure: A new breakthrough in the field of AI GPUs Mass production is expected in 2027
The main advantages of FOPLP technology are its low unit cost and large package size. This means that AI GPUs packaged with FOPLP technology will have better price-performance and better performance. In addition, the larger package size also makes it possible to integrate more circuits and functions, further improving the performance of AI GPUs.
As an innovative achievement in the field of semiconductor packaging, FOPLP technology originates from the continuous exploration of packaging technology by leading companies such as TSMC. Compared to traditional packaging methods, FOPLP technology significantly improves the number of monolithic outputs and cost-effectiveness by replacing IC substrates with large-area panels. This technology not only reduces the unit cost, but also makes the package design more flexible and diverse, meeting the demanding requirements of high-performance chips such as AI GPUs for package size and transmission efficiency.
According to a research report by TrendForce, since the second quarter of 2024, AMD and other chip companies have been actively approaching TSMC and OSAT (Professional Packaging and Testing Foundry) companies to discuss the use of FOPLP technology for chip packaging. The attention of these companies to FOPLP technology indicates that the packaging industry will usher in a major technological innovation.
Although the application of FOPLP technology in the field of consumer ICs may reach mass production earlier, the mass production time of the AI GPU field is expected to be in 2027. This estimate is based on the current level of development of FOPLP technology and industry needs, while also taking into account technical challenges and uncertainties in the development of equipment systems.
