SK Hynix, the world’s second-largest memory chip maker and a key Nvidia supplier, will invest 9.4 trillion won ($6.8 billion) for its inaugural chip plant in South Korea. Kim Young-sik, the company’s head of manufacturing technology, explained that this is a strategic investment for the company in response to the surge in demand for AI semiconductors. The ambitious project will involve building four state-of-the-art semiconductor plants near Seoul. The construction is expected to start in March next year, and its completion is slated for May 2027.
The site will span 4.2 million square meters and will house four cutting-edge chip plants and over 50 local firms in the semiconductor sector. The facility will also boast a ‘mini-fab’ research centre for processing 300-mm silicon wafers, offering local chip materials and equipment manufacturers a realistic environment to test their innovations.
Why does it matter?
It is worth noting that this new fab will be set in the Yongin Semiconductor Cluster near Seoul, where the government aims to build a large-scale chip operations complex. As such, SK Hynix’s investment will help supplement the South Korean government’s efforts to sustain its leadership in-memory technology, which is crucial for AI applications.
Britain has initiated a new technology security partnership with India, aiming to boost economic growth and collaboration in telecom security while fostering investment in emerging technologies. The agreement will enhance cooperation on critical technologies, including semiconductors, quantum computing, and AI.
British Foreign Secretary David Lammy emphasised that this partnership would address future AI and critical minerals challenges, promoting mutual growth, innovation, job creation, and investment. Lammy made these remarks during his visit to India, where he met with Prime Minister Narendra Modi and India’s Minister for External Affairs.
Additionally, both nations have committed to closer collaboration on tackling climate change. That includes mobilising finance and advancing partnerships in offshore wind energy and green hydrogen.
Samsung’s high bandwidth memory chips HBM3 have been approved by Nvidia for use in its AI processors, specifically for the H20 chip developed for the Chinese market, in compliance with the US export controls. Samsung may begin supplying these chips to Nvidia starting in August.
Despite Samsung being one of the world’s largest memory chip manufacturers, it still needed help to get Nvidia to certify its HBM chips. It is still being determined if Nvidia will use Samsung’s HBM3 chips in its other AI processors or if further testing is required. Meanwhile, Samsung’s fifth-generation HBM3E chips are still being tested to meet Nvidia’s standards.
Why does it matter?
AI chips require large amounts of high-speed memory, and HBM is a type of dynamic random access memory (DRAM) with a uniquely stacked memory chip design, which provides the necessary speed and capacity. Although HBM was introduced in 2013, its demand has risen drastically with the AI boom in recent years. Currently, only Micron, Samsung, SK Hynix manufacture HBM chips, and Nvidia have already certified HBM3 chips from Micron and SK Hynix, but there remains a shortage. Thus, Nvidia’s decision to approve Samsung’s HBM3 chips enables Nvidia to expand its supply chain and meet the deficit demand.
The movement of illicit semiconductor chips to Russia has seen a notable decrease, according to the US Department of Commerce, driven by intensified efforts from Western nations to curb the flow of technology that could support Moscow’s military capabilities. However, China and Hong Kong continue to serve as significant transhipment hubs, enabling the ongoing, albeit reduced, supply of these critical components.
Western governments have ramped up measures to control the export of semiconductor chips and other sensitive technologies to Russia. These actions are part of broader sanctions aimed at restricting Russia’s access to advanced technology that could be used in military applications. Enhanced scrutiny and stricter enforcement of export controls have been pivotal in reducing the volume of chips reaching Russia.
According to the Semiconductor Industry Association (SIA), the volume of semiconductor chips illicitly reaching Russia has fallen by approximately 20% over the past six months. Despite the slowdown, China and Hong Kong remain key transhipment hubs for semiconductor chips destined for Russia. Playing a crucial role in the supply chain, they often serve as intermediary points where chips are re-exported to avoid detection.
China and Hong Kong are employing sophisticated methods to circumvent sanctions and export controls, including re-routing shipments, re-labelling products, using shell companies, misdeclaring goods, and complex supply chains.
Why does this matter?
Industry experts highlight that while the reduction in chip flows is a positive development, the continued role of China and Hong Kong as transhipment hubs poses ongoing challenges. The complex nature of global supply chains and the high demand for semiconductors make it challenging to eliminate illicit flows entirely.
According to sources, Nvidia is developing a version of its flagship AI chips for the Chinese market to comply with US export controls. The new chip, part of the ‘Blackwell’ series unveiled in March, is expected to be produced later this year. The ‘B200’ model in this series significantly outperforms its predecessor in tasks like chatbot responses. Nvidia is collaborating with its major Chinese distributor, Inspur, to launch this chip, tentatively named the ‘B20.’
In response to tighter US export controls introduced in 2023 to prevent advancements in Chinese supercomputing, Nvidia has created three chips specifically for China. Despite initial setbacks with the H20 chip, sales have surged, with projections of over 1 million units sold in China this year, amounting to over $12 billion. These developments highlight Nvidia’s strategic efforts to maintain its market presence amidst growing competition from Chinese firms like Huawei.
As the US continues to enforce and potentially expand semiconductor-related export restrictions, Nvidia’s actions reflect the broader impact on the global chip industry. The Biden administration’s plans to implement stricter controls on AI technology and efforts to influence policies in other major tech-producing countries underscore the ongoing geopolitical tensions in the semiconductor sector.
Taiwan Semiconductor Manufacturing Co (TSMC), the leading producer of advanced chips for AI applications, is set to report a 30% rise in second-quarter profit on Thursday, driven by soaring demand. The world’s largest contract chipmaker, which counts Apple and Nvidia among its clients, has seen its stock—and the broader Taiwan market—reach record highs. Last week, TSMC’s American Depositary Receipts surpassed a trillion-dollar market value.
For the quarter ending 30 June, TSMC is expected to announce a net profit of T$236.1 billion ($7.25 billion), a significant increase from the T$181.8 billion reported in the same period last year. The company recently reported a substantial rise in second-quarter revenue, exceeding market expectations. Analysts like Li Fang-kuo from President Capital Management Co. anticipate a strong third-quarter outlook for all of TSMC’s products.
TSMC is heavily investing in expanding its production capacity, including spending $65 billion on three new plants in Arizona. However, the majority of its manufacturing will remain in Taiwan. The company is expected to maintain its capital expenditure guidance for this year at $28 billion to $32 billion, with 70% to 80% allocated to advanced technologies. KGI Securities’ Chu Yen-min suggests TSMC raise its capital spending due to favourable market conditions, which could further boost its stock price and support the broader market.
Why does this matter?
TSMC, often called the ‘sacred mountain protecting the country’ for its crucial role in Taiwan’s economy, remains a dominant player in the semiconductor industry despite challenges from Intel and Samsung. The AI boom has significantly increased TSMC’s stock price, which has surged 75% this year, outpacing the 33% gain in the broader market.
SoftBank Group, the Japanese multinational investment holding company, has acquired Graphcore, a British AI chipmaker, in a strategic business move that ends speculation about Graphcore’s future amid financial struggles. Once positioned as a competitor to Nvidia, Graphcore has faced challenges securing sufficient investment despite its technology potential.
Graphcore, valued at $2.77 billion in 2020, had been grappling with financial viability, including layoffs and operational closures. CEO Nigel Toon acknowledged the company’s difficulties but expressed optimism about the deal with SoftBank, highlighting the substantial resources it brings.
Toon emphasised the significant investment from SoftBank, noting its transformative impact on Graphcore’s global competitiveness. However, he pointed out structural barriers in the UK tech industry, such as limited domestic investment from pension funds, hindering growth opportunities.
Regarding potential collaboration with SoftBank-owned Arm Holdings, a leading chip designer, Toon indicated Graphcore’s intention to leverage synergies within SoftBank’s portfolio, although specifics were not disclosed.
The Biden administration announced a plan to allocate up to $1.6 billion to advance technology for packaging computer chips, a crucial step to maintain the US edge over China in fields like AI. The funding, authorised under the 2022 CHIPS Act, aims to innovate faster data transfer methods between chips and manage their heat generation. Laurie Locascio, an under-secretary in the Commerce Department, announced the initiative at an industry conference, signalling companies to apply for grants up to $150 million each.
The CHIPS Act, which received bipartisan support, allocates $52 billion to bolster domestic chip production, mainly focusing on factories that transform silicon wafers into chips. The US currently contributes about 10% to this industry, with much of the activity outsourced to Asia. Packaging, an essential process that attaches finished chips to substrates, is primarily done in Taiwan, Malaysia, South Korea, the Philippines, Vietnam, and China. The US handles only about 3% of advanced chip packaging.
Why does it matter?
Federal funding will target the next stages of chip production, ensuring chips made in the US can be sent to somewhere other than Asia for packaging to reduce dependence on foreign companies. The shift aligns with the industry’s push for perfecting computing performance by combining multiple chips. Companies like Nvidia and Intel are already making strides in this area, with federal support helping to keep US firms at the forefront of technology. The new grants are part of a $3 billion initiative under the National Advanced Packaging Manufacturing Program, aiming to foster innovation and self-sufficiency in the semiconductor sector.
At the recent World Artificial Intelligence Conference in Shanghai, Chinese GPU developers seized the opportunity to showcase their products in Nvidia’s absence. Prominent companies such as Iluvatar Corex, Moore Threads, Enflame Technology, Sophgo, and Huawei’s Ascend were at the forefront, highlighting their advancements despite significant challenges in manufacturing and software ecosystems.
Enflame Technology emphasised the shift from foreign-dominated computing clusters to a mix of Chinese and foreign GPUs. The company, along with AI solutions firm Infinigence, is promoting compute resources that utilise a variety of chips from both Nvidia and Chinese manufacturers. However, US export restrictions have prevented Nvidia from selling its most advanced chips in China, and several Chinese firms, including Huawei, are struggling with manufacturing hurdles due to being blacklisted by the US.
Huawei’s booth was a major attraction, showcasing its Ascend 910B chips, which train numerous large language models in China. Meanwhile, Enflame presented its Cloudblazer T20 and T21 AI-training chips, benefiting from not being on the US trade blacklist, which allows it access to global foundries like TSMC.
Despite these efforts, Chinese GPUs still need to catch up with their global counterparts regarding performance. Nvidia remains a dominant player, with tailored chips for the Chinese market continuing to be popular. Nvidia is expected to deliver over 1 million H20 GPUs in China this year, generating $12 billion in sales. However, experts highlight that China’s in-house technology still needs to meet its substantial domestic AI demand.
Elon Musk’s AI startup xAI and Oracle have ended discussions on a potential $10 billion server deal. The talks aimed to expand an existing agreement where xAI rents Nvidia AI chips from Oracle. Musk stated that xAI would build a system independently using Nvidia’s H100 graphics processing units for quicker completion.
A source revealed that the specific capacity discussed had been allocated to another customer. Despite this, Oracle continues to engage with xAI regarding its infrastructure needs. Issues such as Musk’s ambitious timeline for building a supercomputer and concerns about power supply at the preferred location contributed to the breakdown of talks.
xAI still contracts with Oracle to train AI models in Oracle’s Gen2 Cloud.
