TSMC to use GlobalFoundries Malta fab for AI packaging

On 9 October 2026, heise online reported that TSMC has agreed to manufacture silicon interposers at GlobalFoundries' facility in Malta, New York.

The five-year deal is valued at $2 billion and targets mass production in the first half of 2028. This partnership allows TSMC to expand its advanced packaging capacity in the United States while bringing GlobalFoundries, typically known for automotive and connectivity chips, into the CoWoS ecosystem used by Nvidia and AMD. The move signals a significant shift in how advanced packaging is handled across the Atlantic, as TSMC seeks to reduce its reliance on Asian facilities for critical AI components. By embedding itself in GlobalFoundries' US operations, TSMC is effectively creating a parallel supply chain that can respond to domestic demand without the logistical friction of international shipping. This strategic alignment also positions GlobalFoundries as a key player in the high-performance computing space, a sector where it has historically played a secondary role compared to its automotive and industrial focus. The financial commitment of $2 billion highlights the scale of the investment required to compete in the advanced packaging market, where yield rates and thermal management are critical to performance. Analysts suggest that this deal could reshape the competitive landscape for other foundries seeking to enter the US market, as TSMC's presence will likely set new standards for quality and efficiency in the region. The timing of the announcement, coinciding with heightened geopolitical tensions, adds another layer of complexity to the global chip supply chain, where national security concerns are increasingly driving commercial decisions.

The interposers act as the data connection layer for large chips, combining processors and memory on a single silicon base. According to heise, this technology is critical for AI accelerators like Nvidia's Blackwell and AMD's Instinct-MI series. GlobalFoundries vice president Ed Kaste stated that advanced packaging is increasingly decisive for next-generation AI system performance and scalability.

The stock price of GlobalFoundries rose by roughly 7 percent on Thursday following the announcement, adding to a nearly 30 percent gain since the start of the year.

Memory and foundry ties tighten

Supply chain security is not limited to packaging. On 8 October, Tom's Hardware reported that AMD CEO Lisa Su is in South Korea to secure additional High Bandwidth Memory (HBM) supply from Samsung. Su described Samsung and SK hynix as key supply-chain partners, noting that AI demand remains extremely high. The report, citing Chosun.biz, claims Samsung hopes to convert its memory relationship with AMD into foundry orders for logic chips, broadening the partnership beyond data storage components. This expansion of the partnership reflects the growing interdependence between memory manufacturers and logic foundries, as the demand for high-performance computing continues to outpace supply. Samsung's push to secure foundry orders is a strategic move to diversify its revenue streams and reduce its exposure to the volatile memory market, which has seen significant price fluctuations in recent years. By deepening its ties with AMD, Samsung is positioning itself as a key supplier for the next generation of AI chips, which require both advanced memory and high-performance logic. This move also highlights the competitive pressure on SK hynix, which has traditionally been the dominant supplier of HBM for AMD. The potential shift in the memory supply chain could have ripple effects across the industry, as other manufacturers seek to secure their own supply of HBM for their AI products. The outcome of these negotiations will likely shape the competitive dynamics in the AI chip market for years to come.

These moves occur against a backdrop of intensifying geopolitical friction. Tom's Hardware also reported on 8 October that Taiwanese prosecutors indicted 10 company leaders for smuggling US military-grade chips from Texas Instruments and Analog Devices to China. The chips, which included radar receivers and electronic reconnaissance gear, were allegedly routed to Chinese missile and radar programs using forged orders from Taiwan's National Chung-Shan Institute of Science and Technology. The US Commerce Department prohibits these specific high-speed parts from being shipped to China, highlighting the strict enforcement of export controls that shapes global chip logistics. This case illustrates the risks associated with the global chip supply chain, where even a single point of failure can have significant implications for national security. The indictment of 10 company leaders signals a heightened level of enforcement by Taiwanese authorities, who are increasingly aware of the potential for their country to be used as a conduit for smuggling sensitive technology. This move also highlights the complexity of the global chip supply chain, where components are often sourced from multiple countries and passed through several intermediaries before reaching their final destination. The case serves as a reminder of the importance of maintaining strong export controls and ensuring that sensitive technology does not fall into the wrong hands. The outcome of this investigation could have significant implications for the future of the global chip supply chain, as it highlights the need for greater transparency and accountability in the industry.

Grid power and automotive impact

The rapid expansion of AI infrastructure is also straining power grids, which indirectly affects chip design and supply. EE Times reported on 7 October that Axiado CEO Gopi Sirineni argued power, not chips, is the limiting factor in AI buildout. Data center racks are moving from 5-10 kW to 240 kW with the Vera Rubin generation, necessitating dynamic voltage and frequency scaling to manage heat and energy spikes. This infrastructure bottleneck may influence how automakers plan their own data centers for autonomous driving development, though direct automotive production remains focused on different chip types. The shift to higher power consumption is driven by the increasing complexity of AI models, which require more computational power to process larger datasets and perform more complex tasks. This trend is putting significant pressure on power grids, which are already struggling to meet the demands of traditional data centers. The move to 240 kW racks represents a significant leap in power consumption, requiring substantial upgrades to existing infrastructure. This shift is also driving innovation in power management techniques, as manufacturers seek to reduce the energy consumption of their data centers. The impact of this trend on the automotive industry is likely to be indirect, as automakers are focused on developing their own data centers for autonomous driving development. However, the need for efficient power management is likely to influence the design of automotive chips, as manufacturers seek to reduce the energy consumption of their vehicles. This trend is likely to drive further innovation in the automotive sector, as manufacturers seek to develop more efficient and sustainable power management solutions.

Automakers are feeling the pinch of these complex supply chains in a different way. Ford CEO Jim Farley stated on 3 October that Europe is effectively lost to Chinese electric vehicles, citing data from Schmidt Automotive Research showing that one in seven battery EVs sold in Western Europe in the first five months of 2026 belonged to Chinese brands. While Farley warned Washington to be considerate before allowing Chinese EVs into the US, the underlying issue is the speed of Chinese manufacturing and component integration. Ford's own partnerships with Geely and CATL have drawn scrutiny from the US Transportation Secretary, who warned that such ties challenge national security and supply-chain independence policies. The rapid growth of Chinese EVs in Europe demonstrates the strength of their manufacturing capabilities and the efficiency of their supply chains. This trend is likely to continue, as Chinese manufacturers continue to invest in R&D and expand their production capacity. The impact of this trend on the US market is likely to be significant, as Chinese EVs are likely to be more competitive in terms of price and features. This trend is also driving innovation in the US automotive sector, as manufacturers seek to develop their own competitive EVs. The outcome of this competition will likely shape the direction of the automotive industry, as manufacturers seek to develop more efficient and sustainable power management solutions.

Meanwhile, Rohm Semiconductor is expanding its back-end manufacturing in India to mitigate risk. EE Times reported on 8 October that Rohm has partnered with Tata Electronics and Suchi Semicon for assembly and test operations. Managing director Makoto Terada explained that the company is keeping wafer fabrication in-house but outsourcing packaging for existing legacy products like power MOSFETs. This strategy allows Rohm to benchmark Indian OSAT capabilities before potentially expanding to global production, a move that reflects the broader trend of diversifying supply chains away from single-region dependencies. The expansion of Rohm's back-end manufacturing in India is a strategic move to reduce its reliance on Asian facilities and mitigate the risks associated with geopolitical tensions. This move also highlights the growing importance of India as a manufacturing hub for the global chip industry, as the country continues to invest in its semiconductor ecosystem. The partnership with Tata Electronics and Suchi Semicon is a significant step towards establishing a resilient supply chain in India, which could potentially serve as a model for other manufacturers seeking to diversify their supply chains. The outcome of this partnership will likely shape the direction of the global chip supply chain, as it highlights the need for greater diversification and resilience in the industry.

The situation for automakers is further complicated by the need for specialized chips. MintNeuro, an Imperial College London spinout, raised $5 million on 7 October to scale chips purpose-built for neural devices, including potential applications in automotive health monitoring and driver assistance. While not directly for vehicle control, such specialized silicon highlights the fragmentation of the chip market, where general-purpose processors no longer cover all emerging automotive and AI applications. The rise of specialized chips is a response to the growing demand for more efficient and powerful computing solutions, as traditional general-purpose processors struggle to meet the demands of emerging applications. This trend is likely to drive further innovation in the chip industry, as manufacturers seek to develop more specialized and efficient computing solutions. The impact of this trend on the automotive industry is likely to be significant, as automakers seek to develop more advanced and efficient vehicle systems. The outcome of this trend will likely shape the direction of the automotive industry, as manufacturers seek to develop more efficient and sustainable power management solutions.

As TSMC and GlobalFoundries prepare for the 2028 production run, and as AMD secures memory deals in Asia, the chip industry is visibly reorganizing. The integration of foundries, memory providers, and packaging specialists into tighter, region-specific alliances is becoming the new norm. For automakers, this means navigating a supply chain that is increasingly fragmented by both technology type and geopolitical alignment, requiring long-term partnerships rather than spot purchases to secure the components needed for the next generation of vehicles. The reorganization of the chip industry is a response to the growing complexity of the global supply chain, where geopolitical tensions and technological advancements are driving a shift towards more localized and resilient supply chains. This trend is likely to continue, as manufacturers seek to reduce their reliance on single-region dependencies and mitigate the risks associated with geopolitical tensions. The impact of this trend on the automotive industry is likely to be significant, as automakers seek to develop more advanced and efficient vehicle systems. The outcome of this trend will likely shape the direction of the automotive industry, as manufacturers seek to develop more efficient and sustainable power management solutions.