Foundry & Contract Fabrication

Intel Corporation's Arizona chip factory venture has begun its production contract, binding the company to minimum output and inventory limits that can trigger volume-related damages payable to the venture. The filing for the quarter ended June 27, 2026, discloses the arrangement, under which falling short of minimum production levels or exceeding on-site inventory limits carries contractual consequences. The partnership dates to 2022, when Intel became the venture's 51% owner and a Brookfield-managed investor held 49%, with the original announcement outlining up to $30 billion of joint investment in two factories in Chandler, Arizona. The venture recorded approximately $4.1 billion in partner contributions during the first half of 2026, and Intel retains the right to purchase all of the venture's wafer output while consolidating its results. The quarterly note does not quantify the minimum production threshold, inventory ceiling, or damages formula, leaving the cost of a demand shortfall uncalculable.

Bank of America initiated coverage of IonQ on September 28 with a Buy rating and a $60 price target, calling the company the largest public pure-play quantum company by revenue. The call lands as IonQ reported $80.1 million of Q2 2026 revenue, up 287% year-over-year, its strongest quarter on record and a fifth consecutive quarter of record results, with organic revenue up 132%. Management raised full-year 2026 IonQ revenue guidance to $280 million to $290 million and maintained its expectation for approximately 100% organic growth, while remaining performance obligations reached $485 million, up from $122 million a year earlier. Following its $1.8 billion acquisition of SkyWater, IonQ said it received its first fully integrated quantum processing units from the foundry and plans to begin commissioning 256-qubit systems in 2027, and it announced Superion 256 deployments at Florida International University and NVIDIA's Accelerated Quantum Research Center. BofA also pointed to IonQ's semiconductor-enabled approach to scaling qubit counts and its expansion across quantum computing, communications and sensing.

Taiwan Semiconductor Manufacturing expanded its chip design partnership with Synopsys, strengthening the tools customers can use to design chips for TSMC's upcoming A14 process. The September 23 announcement covers AI-assisted design, automated chiplet layouts, power analysis for advanced packaging and verification for co-packaged optics, and Synopsys also reported progress on interfaces for HBM4, PCIe 7.0 and Ethernet. TSMC has scheduled A14 volume production for 2028. The U.S.-listed shares fell about 0.9% to $446.57 at 11.48am ET on September 28, a level 22.79% above the $363.69 GF Value. Better design tools could help customers get chips ready sooner, but they do not guarantee orders, and investors still need to see customer designs reach production, strong yields and fabs running at capacity.

Alibaba Group laid out plans for a five trillion to ten trillion parameter model and a new Zhenwu V900 chip at its September 22 conference. The V900 delivers three times the performance of its predecessor and can power clusters of up to 500,000 chips, while CEO Eddie Wu set a target for Alibaba Cloud to operate more than 20 gigawatts of data center capacity by 2032. Reuters reported that the announcement lifted Alibaba shares on September 22, but the U.S. shares slipped about 0.1% to $109.63 at 11.10am ET on September 28, showing that early excitement has since cooled. The investment case rests on turning that computing scale into paying cloud demand, since building its own models and chips could give Alibaba more control over costs but a 20 gigawatt buildout raises the stakes for spending and utilization. The shares sit $11.86 below the $121.49 GF Value estimate, a 9.76% gap that will matter only if the expansion earns an adequate return.

SK hynix is in exploratory discussions with Intel over a potential manufacturing partnership in the United States that could see memory chips produced at Intel's planned Ohio complex. Media reports indicate one option involves SK hynix leasing part of the facility, while another could involve a joint venture including major cloud companies, though SK hynix said no specific arrangement has been decided. The talks come as SK hynix builds out its US footprint, having recently broken ground on its more than $4 billion Indiana facility, which is expected to begin producing next-generation high-bandwidth memory products in the second half of 2029, with wafers made in South Korea slated for advanced packaging and testing there. SK hynix reported record second-quarter 2026 results, with revenues surging 257% year over year and operating profit jumping 557%, and has established long-term agreements with around 10 key customers while ramping HBM4. SKHY has gained 25.8% over the past three months against the industry's decline of 12.6%, outperforming FormFactor, Inc., which has lost 10%, and ASE Technology Holding Co., Ltd., which is up 5.2% over the same period. SKHY currently carries a Zacks Rank #2 (Buy).

United Nova Technology disclosed an earnings forecast on the evening of September 28, expecting to turn from loss to profit in the first three quarters of 2026. The company expects operating revenue for the first three quarters to be about 7.766 billion yuan, an increase of about 2.344 billion yuan from the same period last year, up about 43.23 percent year on year. Net profit attributable to owners of the parent is expected to be about 681 million yuan, an increase of about 1.144 billion yuan from the same period last year, up about 247.08 percent year on year. The company said that during the reporting period, driven by multiple factors including global semiconductor industry technology iteration, demand upgrades and scenario penetration, as well as the continued expansion of emerging application areas such as AI, the company actively expanded its market, deepened customer cooperation, seized opportunities in domestic substitution, maintained a high level of capacity utilization, continuously optimized its product structure, and the scale effect released by sales volume growth together with cost control supported profitability. United Nova Technology was founded in 2018 and listed in May 2023. Since 2019, its performance had been in continuous loss, with a net loss attributable to the parent of 1.958 billion yuan in the year of listing. In the first half of this year, the company's net profit attributable to the parent had already turned from loss to profit, rising 263.40 percent from the same period last year to 278.3 million yuan. The company noted that the above forecast data are only preliminary calculations, and the specific accurate financial data are subject to the officially disclosed third quarter report for 2026.

Alibaba Group Holding Limited is deepening its full-stack AI strategy with a planned next-generation model containing 5 trillion to 10 trillion parameters, as much as four times the size of its current flagship Qwen 3.8 Max, alongside the unveiling of the Zhenwu V900 AI chip, which delivers three times the performance of its predecessor and is designed to operate in large clusters. Mass production of the chip is expected to begin in early 2027, and Alibaba also aims to expand Alibaba Cloud's data-center capacity to more than 20 gigawatts by 2032. Its shares rose 5.1% following the announcements. In the June quarter, Alibaba's AI cloud and compute revenue rose 45% year over year to RMB48.44 billion, or $7.14 billion, while AI-related product revenue reached RMB12.38 billion and maintained triple-digit growth for the 12th consecutive quarter, and cloud adjusted EBITA increased 133% to RMB5.94 billion, lifting the margin to about 12%. Alibaba spent RMB67.68 billion, or $9.98 billion, on capital expenditures in the June quarter, up 75% year over year, largely because of AI infrastructure investment, and quarterly net profit fell 75% as AI-related spending increased. The company expects AI investment to break even within three years.

Alibaba Group Holding Limited unveiled the Zhenwu V900 on September 22, an AI chip the company describes as China's most powerful and roughly three times the performance of its predecessor. The chip is expected to enter mass production with commercial availability potentially beginning in the first quarter of next year, and Alibaba says its existing Zhenwu chips are already used by more than 650 customers across automotive, financial services, energy, and manufacturing. Alibaba has committed at least RMB 380 billion, or about $53 billion, over three years to AI and cloud infrastructure, part of which aims to expand its cloud data center capacity to more than 20 gigawatts by 2032. The company cautioned that the threefold performance improvement is measured against the M890 chip currently in use rather than Nvidia's latest-generation accelerators, so it does not establish that Zhenwu has closed the gap with Nvidia's most advanced hardware. Insider Monkey's hedge fund database shows the number of hedge funds holding Alibaba declined to 97 in the second quarter from 102 in the first quarter, while short interest stood at approximately 1.68% of shares, or about 41.04 million shares, as of September 15.

President Donald Trump and President Xi Jinping held a joint summit at the White House, discussing artificial intelligence, security and technology. It marked the first joint statement between the two leaders, amid concerns among executives at major technology companies and some U.S. policymakers that AI could pose a threat to humanity. Trump affirmed that AI development should proceed freely in order to preserve the United States' advantage, while China criticised U.S. semiconductor export controls as creating restrictions on Chinese companies. Xi Jinping said China and the United States have both the capability and the responsibility to develop and manage AI for the benefit of humanity, and that it should always remain under human control. He also called on Trump to change the U.S. position on Taiwan by opposing Taiwan independence. Trump said the two sides had made substantial progress on points of disagreement, referring to efforts to expand access to the Chinese market for U.S. farmers and ranchers under the trade committee mechanism, and he did not mention the Taiwan issue. After the meeting, China announced that it had reached a new joint agreement with the United States, according to Xinhua News Agency, without disclosing details. A few hours earlier, the United States had announced an extension of the trade truce with China by another two months, until January 10. Xi Jinping also said there is potential to add direct flights between the two countries, and announced that two pandas will arrive at Zoo Atlanta in the U.S. state of Georgia within a few days.

ASML Holding won broader customer commitments for its roughly $400 million High-NA EUV machines, even as its U.S. shares fell about 1.8% to $1,713.41 in Thursday morning trading. Intel is already using High-NA in volume production on selected chip layers, Samsung plans to bring the technology into DRAM production in 2028, and TSMC targets advanced chip production from 2030. ASML says its EXE:5200B can print features 1.7 times smaller than its current NXE systems in a single exposure, potentially sparing customers costly manufacturing steps. The catch is price: chipmakers need enough output and yield gains to earn back the cost of each machine, and ASML needs those gains to turn early commitments into lasting demand. At $1,713.41, the shares sit 35.5% above the roughly $1,260 GF Value estimate, a premium that puts pressure on High-NA to deliver on the factory floor.

The National Semiconductor and Advanced Electronics Industry Policy Committee, chaired by Deputy Prime Minister and Finance Minister Ekniti Nitithanprapas, has approved the first national strategy for developing the semiconductor and advanced electronics industry, as disclosed by Narit Therdsteerasukdi, Secretary-General of the Board of Investment, or BOI. The strategy sets out a three-phase roadmap toward the goal of Made-in-Thailand chips, targeting more than 500 billion baht in investment by 2030 and 2.5 trillion baht by 2050, while pushing industry revenue to 5 trillion baht per year and creating more than 230,000 new jobs. The first phase, by 2030, will upgrade the existing chip assembly and testing base to advanced packaging and lay the foundation for upstream chip production. The second phase, by 2040, focuses on attracting major investors in chip design and wafer fabrication. The third phase, by 2050, aims for Thailand to have a complete semiconductor supply chain and to build Thai companies into local champions. It will focus on three main semiconductor groups: photonics, power semiconductors and sensors, driven through five main mechanisms: incentives and tailored investment support packages, high-skilled workforce development, technology and research infrastructure, readiness in electricity, water and clean energy, and regulatory improvement. Meanwhile, the Siam Silica plan aims to produce and develop a total workforce of 86,600 people by 2030, comprising 84,900 high-skilled personnel and about 1,700 senior research personnel. Infineon Technologies, a major chipmaker from Germany, is preparing to open its first plant in Thailand in Samut Prakan province on October 1, 2026, as a base for advanced power module production along with a research and development center, and when fully operational the project is expected to create more than 5,000 jobs for Thai personnel. In addition, several global technology companies have chosen Thailand as a production base, such as Analog Devices, Foxsemicon, ZDT, Terahop and MPI. BOI data shows that from 2023 through June 2026, there were 879 projects applying for investment promotion in the semiconductor and advanced electronics industry, with investment value of more than 909 billion baht.

Citi has raised its spending assumptions for Alibaba and maintained its $190 price target and Buy rating on the stock, according to an article first reported by GuruFocus. The bank's updated forecasts follow Alibaba's push to expand data-center capacity to more than 20 gigawatts by 2032. For fiscal 2027 through fiscal 2029, Citi now projects capital expenditures of 258 billion yuan, 283 billion yuan and 282 billion yuan, respectively, reflecting the expected cost of building out computing infrastructure to support Alibaba's growing AI operations. Citi analyst Alicia Yap estimates Alibaba's external AI cloud revenue could reach $168 billion by fiscal 2033, a forecast implying a 40% compound annual growth rate from fiscal 2026 and dependent on continued expansion of AI computing capacity. Alibaba has also introduced the Zhenwu V900 accelerator and outlined plans for much larger AI models, with Reuters reporting the new accelerator is expected to enter mass production in early 2027.

SK Hynix is exploring what would be its first-ever memory chip manufacturing in the United States, holding talks with Intel Corporation about leasing part of Intel's long-delayed Ohio complex or forming a venture involving Intel and large cloud customers, Reuters reported on September 16, though SK Hynix said no specific plan has been finalized. Intel has committed enormous capital to Ohio, with the broader project potentially reaching around $100 billion over time, and delays have pushed planned production into the next decade, so a tenant or joint-venture partner could help Intel share the financial burden of the buildout. The talks matter for Micron Technology, Inc. because one of its strongest strategic advantages is being the major U.S.-based memory producer while AI has made DRAM and HBM increasingly scarce, and a successful SK Hynix move could eventually erode part of that U.S.-supply differentiation by giving U.S. cloud companies another potential domestic supplier. Hedge fund holdings rose for both chipmakers in the second quarter, with 138 funds holding Intel, up from 112 in the first quarter, and 184 funds holding Micron, up from 154, while Intel had about 152.2 million shares sold short as of August 31, roughly 3% of public float and 1.7 days to cover. The talks do not rescue Intel's foundry strategy or weaken Micron overnight, but they show memory manufacturing becoming strategic enough that SK Hynix is considering advanced capacity on American soil.

The United States and China are competing for dominance in the race to spread artificial intelligence, accelerating their efforts to build self-centered "AI economic spheres." Late last year, the United States launched Pax Silica, an initiative to build secure supply chains for AI, semiconductors, and critical minerals, and the number of signatory countries and regions, centered on allies and friendly nations such as Japan and the United Kingdom, has grown to 25. In July of this year, China also established the World AI Cooperation Organization, or WAICO, with 29 countries to extend cheap AI models and communications infrastructure across the Global South. Dubbed an "AI version of the Belt and Road," it now has 38 participating countries. The US State Department was reported to have drafted a document asking Pax Silica signatories not to join WAICO, pressuring them to take a loyalty test, but Kazakhstan has joined both, and Uzbekistan, Singapore, and Bangladesh are also exploring "dual participation." The US-China summit on the 24th is likely to be a venue for gauging the future of cooperation and division over AI.

Intel shares gained approximately 0.5% to $122.46 as Meta's Muse agent revived the case for server CPUs in an AI market dominated by accelerators. The Muse agent runs through its own virtual computer and can keep working on longer tasks even after a user exits the application, creating demand for general-purpose compute around the model itself. Intel is trying to capture that layer as its business rebounds, with second-quarter revenue rising 25% to $16.1 billion and the company rolling out its Xeon 6+ server processor built on Intel 18A. Still, Muse is evidence of a potentially expanding workload, not evidence of an Intel design win, and Intel now trades 281.26% above its $32.12 GF Value estimate. The real payoff comes only if autonomous agents drive incremental server deployments and Intel wins enough of that spending at attractive margins, making Xeon availability, power efficiency, hyperscaler adoption and Intel 18A execution the numbers investors should watch.