According to a report from the Financial Times, SoftBank Group has decided to move away from its plan to collaborate with Intel on producing AI chips to compete with NVIDIA and is now reportedly focusing on discussions with TSMC.

The same report, citing sources, reported that the partnership between SoftBank and Intel fell through because Intel struggled to meet SoftBank’s requirements. SoftBank reportedly attributed the collapsed talk to Intel’s inability to meet their demands for production volume and speed.

The report noted as well that this fallout occurred before Intel’s announcement of releasing its official announcement on its Q2 (April-June) earnings in early August. Notably, in response to a significant drop in its performance, Intel planned to lay off about 15,000 employees and suspend shareholder dividends.

Moreover, the report further cited rumors claiming that SoftBank has shifted its focus to discussions with TSMC; however, no agreement has been reached so far.

Reportedly, Intel, SoftBank and TSMC have all declined to comment on the situation.

Read more

• [News] Intel Has Offloaded Stake in ARM, Reportedly Raising nearly USD 150 Million
• [News] Intel’s Earnings Fall Short, Cutting Over 15% of Workforce

(Photo credit: Intel)

Masahiro Okitsu, President of Sharp, a subsidiary of Foxconn, announced in a joint interview with Japanese media that the Sakai Display Product (SDP) plant, which produces large-size LCD panels for TVs, will cease production earlier than planned. Initially scheduled for the end of September, the shutdown is now expected to occur in late August.

According to a report by Asahi Shimbun on July 16, Sharp President Masahiro Okitsu announced in a joint media interview that the SDP, which produces LCD panels for TVs, is expected to completely cease production after the Obon festival (late August). This interview marks Okitsu’s first media appearance since taking office as Sharp’s president on June 27.

Okitsu mentioned that around July 20, SDP will start put glass substrates into production, with the final batch of LCD panels produced a month later.

Sharp’s SDP land and plant will be transformed into an AI data center, and a collaboration with Softbank and KDDI has been announced. However, Masahiro Okitsu did not elaborate on this cooperation framework during the media interview on July 16th.

Regarding Foxconn Chairman Young Liu’s appointment as Chairman of Sharp, Okitsu stated that this makes the division of duties clearer. Foxconn will oversee and support Sharp, while the existing brand business operations will be managed by Sharp.

Okitsu also pointed out the goal of increasing the operating profit margin of the “brand business,” which includes products like white goods, to 7% by the 2027 fiscal year (compared to less than 5% in the 2023 fiscal year). He emphasized that achieving a return to profitability in the 2024 fiscal year is imperative.

Reportedly, with Foxconn’s technical assistance, Sharp also plans to enter the AI and electric vehicle (EV) businesses, with plans to launch in the 2026-2027 fiscal years.

On May 14, Sharp announced its financial report, indicating that due to impairment losses in its panel business, the net loss for the 2023 fiscal year (April 2023 – March 2024) reached JPY 149.9 billion, marking the second consecutive year with a net loss exceeding 100 billion yen. However, Sharp forecasts that for the 2024 fiscal year (April 2024 – March 2025), its consolidated operating profit will be JPY 10 billion, with a consolidated net profit estimated at JPY 5 billion.

Read more

• [News] Foxconn Ventures into Advanced Packaging, Sharp to Follow with Production Capacity in 2026
• Sharp Announces Closure of Sakai Gen 10 Line, Supply Cut of 5 Million TV Panels Expected in 2025, Says TrendForce

(Photo credit: Sharp)

Recently, Reuters reported that SoftBank Group acquired Graphcore, a company often referred to as the “UK’s NVIDIA,” though the amount of the deal was not disclosed. Graphcore is a startup in the field of artificial intelligence (AI) that has designed a new type of Intelligent Processing Unit (IPU). In certain model tests, its performance has surpassed that of NVIDIA’s GPU systems, and thus, the industry is optimistic about its potential to compete with NVIDIA’s GPU.

• The Differences between IPU and GPU

As a processor specifically designed for AI computation, also known as an AI processor, IPU is excels in fields such as deep learning, machine learning, and natural language processing, boasting the capability of accelerating various AI-related tasks.

GPU, on the other hand, was initially designed to meet the demands of graphics rendering and image processing. With the rapid proliferation of AI and big data technologies, high-performance GPU, known for their powerful parallel processing capabilities, can handle multiple data points and tasks simultaneously, thereby speeding up training and inference processes, which together enable GPU to be gradually applied in the AI field, particularly in deep learning and machine learning.

Although both IPU and GPU can be used in the AI domain, they differ a lot in several aspects, such as computational architecture and memory architecture.

Previously, Lu Tao, the President and General Manager of Greater China at Graphcore, explained that the Graphcore C600 has 1,472 processing cores per IPU, capable of running 8,832 independent program threads in parallel. In comparison, NVIDIA’s GPU SM Core (Stream Multiprocessor Core) has around 100 cores, varying with different product configurations.

In terms of memory architecture, NVIDIA’s GPUs have a two-level memory structure. The first level consists of around 40-50 MB of memory within the chip, with external HBM or VRAM attached. Graphcore’s IPU, however, contains 900 MB of on-chip SRAM storage, which is distributed.

Lu summarized that the IPU’s architecture shows greater advantages for tasks with high requirements for sparsity and high dimensions relative to to GPU. For matrix operations, its performance might be similar to GPU or slightly less competitive.

Read more

• [News] SoftBank Founder Masayoshi Son Plans to Raise USD 100 Billion to Establish AI Chip Company

(Photo credit: Graphcore)

SoftBank Group’s global IP leader, Arm, has reportedly announced the establishment of an AI chip division with the goal of developing AI chip prototypes by spring 2025. According to a report from DRAMeXchange, mass production will be handled by contract manufacturers, with initial production slated to begin in autumn 2025.

Arm will cover the initial development costs estimated to reach several trillion yen, funded by SoftBank Group. Once a large-scale production system is built, Arm’s AI chip business may be spun off and incorporated into a SoftBank Group division, which is because SoftBank holds a total of 90% of Arm’s shares and has been in talks with TSMC to secure production capacity.

Arm is a significant player in the global semiconductor industry, renowned for its energy-efficient Arm architecture, which commands over 90% of the global market share in smartphone chip field. SoftBank acquired Arm in 2016 for USD 32 billion, empowering Arm to go public on the US stock exchange in September 2023.

Last week, Arm reported fiscal 2024 fourth quarter revenue of USD 928 million (+47% YoY) and adjusted operating profit of USD 391 million. It forecasts first-quarter revenue for fiscal year 2025 to be USD 875-925 million, expecting annual revenue to be USD 3.8-4.1 billion.

According to Canada’s Precedence Research, the current market size for AI chips is USD 30 billion, expected to exceed USD 100 billion by 2029 and USD 200 billion by 2032. Despite NVIDIA’s dominant position in AI chip technology, it is unable to meet the growing demand.

Eyeing on the opportunities presented by the AI wave, SoftBank Group founder Masayoshi Son has identified AI as a key focus area for development and is seeking to raise USD 100 billion to found an AI chip company to compete with NVIDIA.

Read more

• [News] SoftBank Founder Masayoshi Son Plans to Raise USD 100 Billion to Establish AI Chip Company

(Photo credit: SoftBank News)

Novatek has unveiled its AI and related domain strategies during its year-end investor conference. According to a report from Taiwanese news outlet Tai Sounds, Novatek anticipates that the demand for AI will drive an enhancement in display specifications and necessitate the integration of edge devices such as security systems.

Transmission interfaces stand as the core focus of Novatek, with subsequent products extending towards high-speed transmission. In both its driver IC and SoC product lines, Novatek maintains customized product lines, with expectations for growth in customized chips this year.

As per sources cited by Taiwanese news outlet Tai Sounds, Novatek is rumored to be venturing into the IP domain and may have opportunities to collaborate with ARM, a subsidiary of SoftBank Group.

Further clarification from the source also indicates that Novatek specializes in customized power management chips for mobile devices, while Realtek offers USB4 hub solutions.

Read more

• [News] Possible 30% Cut in Orders for New iPad Pro, Impacting TSMC, Foxconn, and Novatek
• [News] Taiwan’s Chip Act Takes Effect in February, TSMC to Benefit from Historic Tax Incentives

(Photo credit: Novatek)