--- title: "Taiwan Semiconductor develops AI chip packaging technology to directly compete with Intel" type: "News" locale: "en" url: "https://longbridge.com/en/news/294497324.md" description: "Taiwan Semiconductor is developing advanced chip packaging solutions that benchmark Intel's EMIB technology to address the capacity bottlenecks caused by the surge in AI computing demand and the competitive pressure from Intel. This technology is jointly developed by Taiwan Semiconductor and Unimicron Technology Corporation, aiming to alleviate the current tight situation in advanced packaging capacity and prevent customers from losing orders due to supply chain diversification" datetime: "2026-07-31T10:23:21.000Z" locales: - [zh-CN](https://longbridge.com/zh-CN/news/294497324.md) - [en](https://longbridge.com/en/news/294497324.md) - [zh-HK](https://longbridge.com/zh-HK/news/294497324.md) --- # Taiwan Semiconductor develops AI chip packaging technology to directly compete with Intel According to two people familiar with the project, Taiwan Semiconductor Manufacturing Company (TSMC) is developing an advanced chip packaging technology that is comparable to Intel's existing solutions. This indicates that the world's leading foundry is wary of the pressure from distant competitors. Chip packaging is the final step in chip manufacturing: it assembles multiple independent silicon chips into a whole and completes the interconnection of circuits to achieve collaborative work while also interfacing with the overall hardware. Packaging was once considered a routine process in chip manufacturing, but with the explosive demand for AI computing power, chip manufacturers need to integrate more processors and high-bandwidth memory into the packaging, resulting in larger sizes and more complex structures. Packaging has now become one of the most prominent capacity bottlenecks in the entire industry. Intel's self-developed Embedded Multi-Die Interconnect Bridge (EMIB) technology relies on micro-silicon bridges to achieve high-speed interconnection between processors and memory. This solution has been favored by companies like NVIDIA: it can support larger-scale multi-chip architectures, help create more powerful AI chips, and facilitate supply chain diversification for chip design manufacturers. The information media The Information reported that NVIDIA is evaluating the use of EMIB to create a next-generation packaging processor that integrates four GPUs. Insiders say that TSMC's internal engineers refer to the advanced packaging project as a "similar to EMIB solution" and have begun discussions about this technology with some customers. TSMC is jointly developing this technology with Taiwan's Unimicron Technology Corporation; Unimicron specializes in packaging substrates, which serve as the foundation for chip packaging and are responsible for signal transmission between the chip and the server. TSMC declined to comment, and Unimicron did not respond to interview requests. TSMC President Wei Zhejia admitted during this month's earnings call that the company's existing advanced packaging capacity is extremely tight, which has limited the expansion pace of customers' AI chips. This capacity gap has provided Intel with a rare opportunity to compete for customer resources that heavily rely on TSMC. Even if customers initially only place packaging orders with Intel, it can help the American company complete technology validation and deepen customer ties, with the long-term potential to further compete for chip foundry orders. The situation has dramatically reversed. Intel has been trying for years to win back high-end chip customers. This American chip company once led the world in advanced process chip manufacturing, but multiple production delays allowed TSMC to surpass it, leading many top chip design companies to shift their high-end main products to TSMC for foundry. Currently, top chips from companies like NVIDIA are all manufactured by TSMC; Intel is fully building its foundry business, aiming to compete with TSMC. Advanced packaging has become Intel's second path to regain industry reputation and attract customers. According to The Information, after months of validating Intel's EMIB technology, Google has finalized an order: by 2028, Intel will complete the packaging of over 3 million self-developed AI processors For TSMC, the research and development of EMIB technology is a defensive measure to avoid Intel's entrenched advantages. According to informed sources, this technology is still in the early stages of development, and the timeline for mass production is currently unclear. However, this is enough to prove that the AI boom is forcing this global leading foundry to expand its packaging technology matrix to meet customers' diverse demands for capacity and next-generation chip heterogeneous integration solutions. Currently, AI chips from manufacturers like NVIDIA and Google predominantly use TSMC's **CoWoS (Chip on Wafer on Substrate)** packaging. AI processors need to continuously retrieve massive amounts of data from memory, requiring the processors and memory to be closely arranged with high-speed interconnect channels. TSMC's CoWoS achieves high-speed interconnection through an entire intermediary layer beneath the processor and memory. In contrast, Intel's EMIB takes a different approach: it embeds micro-silicon bridges at points where high-speed interconnection is needed within the packaging substrate, avoiding the need to stack all components on the same intermediary layer, making it more suitable for ultra-large, complex multi-chip packages. This also means that the substrate is the core component of TSMC's EMIB packaging development. Unimicron Technology can assist TSMC in completing substrate design and manufacturing, supporting the silicon bridge structure and enabling signal connectivity for various chip components above ### Related Stocks - [TSM.US](https://longbridge.com/en/quote/TSM.US.md) - [INTC.US](https://longbridge.com/en/quote/INTC.US.md) - [TSMX.US](https://longbridge.com/en/quote/TSMX.US.md) - [TSMG.US](https://longbridge.com/en/quote/TSMG.US.md) - [TSMZ.US](https://longbridge.com/en/quote/TSMZ.US.md) - [TSMU.US](https://longbridge.com/en/quote/TSMU.US.md) - [INTW.US](https://longbridge.com/en/quote/INTW.US.md) - [STSM.US](https://longbridge.com/en/quote/STSM.US.md) - [TMYY.US](https://longbridge.com/en/quote/TMYY.US.md) - [TSMY.US](https://longbridge.com/en/quote/TSMY.US.md) - [LINT.US](https://longbridge.com/en/quote/LINT.US.md) - [INYY.US](https://longbridge.com/en/quote/INYY.US.md) - [SMH.US](https://longbridge.com/en/quote/SMH.US.md) - [SOXX.US](https://longbridge.com/en/quote/SOXX.US.md) - [SOXL.US](https://longbridge.com/en/quote/SOXL.US.md) - [SOXQ.US](https://longbridge.com/en/quote/SOXQ.US.md) - [XSD.US](https://longbridge.com/en/quote/XSD.US.md) - [PSI.US](https://longbridge.com/en/quote/PSI.US.md) - [FTXL.US](https://longbridge.com/en/quote/FTXL.US.md) - [04335.HK](https://longbridge.com/en/quote/04335.HK.md) - [NVDA.US](https://longbridge.com/en/quote/NVDA.US.md) - [GOOGL.US](https://longbridge.com/en/quote/GOOGL.US.md) - [GOOG.US](https://longbridge.com/en/quote/GOOG.US.md) - [NVD.DE](https://longbridge.com/en/quote/NVD.DE.md) ## Related News & Research - [MediaTek eyes bigger share of AI market](https://longbridge.com/en/news/294536060.md) - [TSMC Just Flexed Its Cash and Raised the Stakes in America After Record AI-Driven Earnings](https://longbridge.com/en/news/293778887.md) - [Key facts: Intel 15‑yr revenue high; margins >40%/17%; 14A/18A risk](https://longbridge.com/en/news/293815661.md) - [Powertech says it is on track to ship for AMD soon](https://longbridge.com/en/news/294091001.md) - [TSMC outlook not hit by quake](https://longbridge.com/en/news/294391659.md)