The US Commerce Department has awarded GlobalFoundries a $1.5 billion subsidy to expand semiconductor production in Malta, New York, and Vermont. This follows the company’s $13 billion commitment to bolstering United States manufacturing over the next decade, with a focus on automotive, AI, and aerospace sectors.
The funding will support enhanced technologies at the Malta facility and plans for a new plant aligned with market demand. New York state has pledged an additional $550 million to support the expansion. Commerce Secretary Gina Raimondo emphasised the urgency to finalise similar agreements before the administration ends.
GlobalFoundries CEO Thomas Caulfield highlighted the critical role of US-made chips in economic and national security. The subsidy is part of the $52.7 billion Chips and Science programme, which also allocated major awards to TSMC, Samsung, and Intel.
As the US prepares for Donald Trump’s second term, China is significantly increasing its semiconductor imports from the US, anticipating potential sanctions. In October, China imported $1.11 billion worth of microchips, a 60% rise from the previous year, and has already imported $9.61 billion in the first ten months of 2024, marking a 42.5% year-on-year increase. This surge reflects China’s growing demand for US semiconductors, particularly CPU-based processors and chips for storage and signal amplification, which align with its AI ambitions.
Despite these imports, China faces hurdles in advancing its chip technology. US sanctions have crippled Huawei’s ability to develop competitive AI chips, with the company’s upcoming processors lagging years behind NVIDIA’s offerings. This setback is largely due to restrictions on access to advanced lithography equipment, such as ASML’s EUV tools, essential for creating cutting-edge chips.
Meanwhile, China has been ramping up its chip manufacturing efforts, investing $25 billion in equipment in the first half of 2024, surpassing spending by Korea, Taiwan, and the US. However, as one-third of global semiconductor demand, China’s position remains critical for the industry. The impact of Trump’s potential tech restrictions, whether broad or selective, will likely influence the global semiconductor market, requiring careful balancing of US production and Chinese demand.
Nvidia reported a staggering $19B in net income last quarter but faced questions about sustaining its rapid growth amid shifts in AI development methods. Analysts questioned CEO Jensen Huang on how Nvidia’s position might evolve with trends like ‘test-time scaling,’ a method that enhances AI responses by increasing computing power during inference, the phase when AI generates answers.
Huang described test-time scaling as a groundbreaking development and emphasised Nvidia’s readiness to support it. He noted that while most of the company’s focus remains on pretraining AI models, the growing emphasis on inference could transform the AI landscape. Nvidia’s dominance in pretraining has propelled its stock up 180% this year, but competition in AI inference is heating up, with startups like Groq and Cerebras offering alternative chip solutions.
Despite concerns about diminishing returns from traditional AI scaling, Huang remains optimistic, asserting that foundational AI development continues to advance. He reiterated Nvidia’s advantage as the largest AI inference platform globally, citing the company’s scale and reliability as critical factors in maintaining its edge.
SK Square, the holding company of AI chipmaker SK Hynix, unveiled plans to enhance shareholder value through a share buyback and other measures. The company will repurchase 100 billion won ($71.51 million) worth of shares within three months and cancel them, following a similar cancellation of shares bought in April.
The move comes after London-based hedge fund Palliser Capital proposed strategies to address SK Square’s undervaluation, with the company’s market value currently less than half the $18 billion worth of its 20% stake in SK Hynix. Palliser, which acquired a 1% stake in SK Square this year, has been in talks with the firm to improve shareholder returns.
This initiative aligns with South Korea‘s “Value-Up” program, designed to encourage companies to increase market value. SK Hynix, a key asset of SK Square, recently reported record profits driven by soaring AI chip demand from Nvidia, adding to the company’s potential for growth.
Huawei plans to begin mass-producing its Ascend 910C AI chip in early 2025, despite ongoing struggles to achieve sufficient production yields due to US trade restrictions. The Chinese telecom giant has already sent samples to tech firms and started taking orders for the chip, designed to rival Nvidia’s high-performance processors. The company faces significant challenges, as restrictions on advanced manufacturing technologies have limited its chip-making efficiency.
The Ascend 910C is produced by Semiconductor Manufacturing International Corp (SMIC) using an N+2 process but suffers from a yield of just 20%—far below the 70% required for commercial viability. Previous Huawei processors, including the 910B, achieved yields of around 50%, leading to delays in fulfilling orders from major clients like ByteDance. Washington’s restrictions, which prevent access to critical Dutch lithography equipment, have further constrained China’s ability to produce advanced semiconductors.
Huawei’s reliance on SMIC has been costly, with chips produced on its advanced nodes priced up to 50% higher than alternatives. While the company has sought supplemental production from Taiwan’s TSMC, US authorities have tightened export controls, limiting access to cutting-edge chips and forcing Huawei to prioritise strategic government and corporate orders. The escalating trade tensions underscore the geopolitical struggle between the US and China over technological dominance, with both nations doubling down on policies to secure their interests.
As Beijing pushes for self-reliance in semiconductors, Huawei’s production challenges reflect the broader impact of US restrictions on China’s tech sector. With further curbs on the horizon, Huawei’s success in advancing its AI chips may shape the next phase of the US -China tech rivalry.
California-based AI startup Enfabrica has raised $115M in a funding round to tackle one of the field’s most pressing challenges, enabling vast networks of AI chips to work seamlessly at scale. The company, founded by former engineers from Broadcom and Alphabet, plans to release its new networking chip early next year. This chip aims to enhance efficiency by addressing bottlenecks in how AI computing chips interact with networks, a problem that slows down data processing and wastes resources.
The startup claims its technology can scale AI networks to connect up to 500,000 chips, significantly surpassing the current limit of around 100,000. This breakthrough could speed up the training of larger AI models, reducing time and costs associated with unreliable or inaccurate outcomes. “The attributes of the network, like bandwidth and resiliency, are critical for scaling AI efficiently,” said Enfabrica CEO Rochan Sankar.
Investors in the funding round included Spark Capital, Maverick Silicon, and corporate backers like Arm Holdings and Samsung Ventures. Nvidia, an industry leader in AI chips, also participated, signaling strong support for Enfabrica’s mission to optimise AI infrastructure.
Silicon Valley firm d-Matrix has launched its first AI chip, designed to enhance AI services like chatbots and video generators. Early samples are being tested by customers, with full-scale shipments expected next year.
The chip focuses on inference tasks, allowing multiple users to interact simultaneously with AI systems, such as generating or modifying videos. d-Matrix’s innovation aims to complement market leaders like Nvidia by specialising in real-time user requests.
Backed by over $160 million in funding, including from Microsoft‘s venture arm, the company has partnered with Super Micro Computer to offer servers equipped with its chips. CEO Sid Sheth highlights strong demand in video applications for multi-user interactions.
With AI adoption surging, data centers are bracing for a 160% jump in electricity consumption by 2030, driven by the energy demands of GPUs. Sagence AI, a startup led by Vishal Sarin, is addressing this challenge by developing analog chips that promise greater energy efficiency without sacrificing performance.
Unlike traditional digital chips, Sagence’s analog designs minimise memory bottlenecks and offer higher data density, making them a viable option for specialised AI applications in servers and mobile devices. While analog chips pose challenges in precision and programming, Sagence aims to complement, not replace, digital solutions, delivering cost-effective and eco-friendly alternatives.
Backed by $58M in funding from investors like TDK Ventures and New Science Ventures, Sagence plans to launch its chips in 2025. As it scales operations, the startup faces stiff competition from industry giants and will need to prove its technology can outperform established systems while maintaining lower energy consumption.
SilTest Semiconductors GmbH and yieldWerx have announced a strategic partnership to optimise European semiconductor yield. That collaboration combines SilTest’s engineering expertise with yieldWerx’s advanced data analytics software, offering semiconductor manufacturers an integrated, data-driven solution for improving production efficiency, optimising yields, and reducing costs.
Specifically, the partnership provides a seamless approach to yield management, addressing critical industry needs by facilitating faster troubleshooting, streamlining testing processes, and offering actionable insights from production data. Furthermore, with SilTest’s presence in Europe, clients benefit from localised support tailored to the specific needs of the European semiconductor market.
In addition, both companies emphasise their commitment to supporting the semiconductor industry with innovative solutions and expert guidance. By merging their strengths, SilTest and yieldWerx aim to empower European semiconductor companies to achieve their yield and efficiency goals, thus driving growth and technological advancement within the industry. Ultimately, this collaboration underscores their dedication to providing cutting-edge tools, region-specific expertise, and deeper insights into yield optimisation, helping clients navigate the challenges of the semiconductor market.
The Korea Fair Trade Commission (FTC) has announced its intention to closely monitor mergers and acquisitions (M&As) in the rapidly growing AI semiconductor sector to assess potential anti-competitive effects. With the increasing pace of technological advancements and the rising demand for AI technologies, the FTC recognises that M&As could significantly impact market competition, particularly by reducing the number of players in the industry.
Given the sector’s dynamic nature and the challenges posed by factors like the US-China tech rivalry, the FTC’s role is crucial in ensuring that these business combinations do not result in unfair market practices, such as price increases or the exclusion of competitors. Therefore, through careful scrutiny of M&As, the FTC aims to preserve healthy competition, support technological innovation, and protect the overall market ecosystem.
In addition, the Korea Fair Trade Commission (FTC) is reviewing several significant mergers, including combining Synopsys and Ansys and acquiring ZT Systems by AMD. These deals could profoundly impact the AI semiconductor market, especially regarding access to essential design software and hardware technologies.
To enhance its understanding of the potential anti-competitive effects of these M&As, the FTC is gathering input from industry experts and academic leaders. This collaborative approach, including forums like the ‘AI Semiconductor-Related Business Combination Forum,’ helps the FTC stay informed about emerging trends and refine its strategies for safeguarding fair competition in the rapidly evolving AI semiconductor sector.
