DELIVERCHIP（HK）LIMITED— 31 July: This Week’s News

This Week’s Highlights

Contents:

I. Intel: Returned to Profit in Q2, but Server Chip Sales Not Expected to Recover Until Q4

II. AMD to Invest US$400 Million in India Over the Next Five Years!

III. Nvidia Launches DGX Cloud Service, Offering Users the Opportunity to Lease AI Supercomputers and the Latest GPUs

IV. Samsung’s Q2 profits plummet by 95 per cent; gradual recovery expected in the second half of the year

V. ASUS unveils AI servers equipped with NVIDIA H100; orders are already booked until next year

Part 1

Intel has released its second-quarter financial results. Revenue stood at US$12.9 billion, down 15 per cent year-on-year; adjusted gross margin was 39.8 per cent, compared with 44.8 per cent in the same period last year; adjusted operating margin stood at 3.5 per cent, compared with 9.3 per cent in the same period last year; adjusted earnings per share were US$0.13, down 54 per cent year-on-year. Net profit was US$1.5 billion, compared with a net loss of US$454 million in the same period last year. Performance across all divisions declined, with Client Computing revenue falling by 12 per cent to US$6.8 billion; Data Centre and AI revenue falling by 15 per cent to US$4.0 billion; the Networking and Edge segment’s revenue fell by 38 per cent to US$1.4 billion; the Autonomous Driving segment’s revenue fell by 1 per cent to US$454 million; whilst the foundry business’s revenue more than tripled to US$232 million. Intel plans to continue making strategic investments in its manufacturing capabilities, having selected Wrocław, Poland, as the location for a new state-of-the-art semiconductor assembly and test facility, with an expected investment of up to US$4.6 billion. Intel forecasts third-quarter revenue of between US$12.9 billion and US$13.9 billion, with an adjusted gross margin of 43 per cent.

Part 2

28 July: According to foreign media reports, AMD’s Chief Technology Officer, Mark Papermaster, announced at the annual semiconductor conference that AMD will invest approximately US$400 million in India over the next five years and establish its largest design centre at the Bangalore Technology Centre.

AMD stated that it will open a new design centre campus in Bangalore by the end of this year and create 3,000 new engineering roles over the next five years.

It is reported that AMD’s new campus will cover 55,555 square feet, bringing the total number of its offices in India to 10; the company currently employs over 6,500 staff in the country.

Although a latecomer to the sector, the Modi government has been actively seeking investment in India’s emerging semiconductor industry. It is worth noting that India launched a US$10 billion incentive scheme for the semiconductor sector in 2021, but to date no company has been granted a licence to establish a manufacturing plant – a key component of the Modi government’s semiconductor strategy.

Insiders indicate that the semiconductor consortium ISMC had planned to invest US$3 billion in a semiconductor plant in India, with Israeli chipmaker Tower as its technology partner; however, the project has been put on hold due to Tower’s ongoing acquisition by Intel. India’s semiconductor market is projected to be worth US$63 billion by 2026.

Part 3

Nvidia recently launched the DGX Cloud service, a platform that allows users to rent AI supercomputers and the latest GPUs. DGX Cloud provides high-performance hardware, including the highly sought-after H100 and A100 GPUs, to meet users’ computing needs in the field of artificial intelligence. Users can choose to rent these systems via Nvidia’s cloud infrastructure or Oracle Cloud Services.

One of the key advantages of DGX Cloud is its ability to meet the multi-node training requirements of large AI models. This is particularly important for projects that require the processing of vast amounts of data and the execution of complex computations. By using DGX Cloud, users can rent high-performance computing resources in the cloud without having to purchase and maintain their own hardware. This not only saves costs but also offers more flexible management of computing resources.

Nvidia’s aim is to operate its AI infrastructure as efficiently as a factory. They seek to treat input data as raw materials and, by running various algorithms and models, transform it into actionable information that companies can utilise. By offering the DGX Cloud service, Nvidia provides users with a more convenient and reliable way to access and utilise these computing resources.

Pricing for DGX Cloud starts at US$36,999 per month, which may be somewhat expensive for individual users. However, for large enterprises and research institutions, this price may be justified, as they gain access to high-performance computing and the latest GPUs without having to invest heavily in purchasing and maintaining hardware.

In summary, Nvidia’s DGX Cloud service offers users a convenient and efficient way to rent AI supercomputers and the latest GPUs. It meets the multi-node training requirements of large AI models and provides users with flexible management of computing resources. Although the pricing may be somewhat expensive for individual users, it represents an attractive option for large enterprises and research institutions.

Part 4

Samsung has released its quarterly financial results, showing a 95 per cent plunge in second-quarter profits. Although global demand for memory chips remains weak, Samsung expects a gradual recovery in the second half of the year, which should drive an improvement in profitability. According to figures released by Samsung, revenue for the second quarter of this year stood at approximately 60.01 trillion won (equivalent to US$47.21 billion), whilst industry analysts had previously forecast 60.8 trillion won. Operating profit for the second quarter stood at approximately 0.67 trillion won (equivalent to US$524 million), whilst the company had previously forecast 0.6 trillion won. Samsung stated that strong demand for artificial intelligence led to DRAM shipments exceeding expectations in the second quarter, thereby improving the performance of its memory business. Looking ahead to the second half of the year, Samsung anticipates a recovery in demand and will focus on high-value-added products, whilst increasing investment in infrastructure, research and development, and packaging technology. Nevertheless, due to persistent macroeconomic challenges such as inflation, overall demand in the mobile phone market has declined compared with the first quarter. Samsung stated that it plans to boost sales of the Galaxy S23 and Galaxy A series. Global smartphone shipments are expected to decline this year. Samsung said it would further reduce production of certain DRAM and NAND products to accelerate inventory normalisation.

Part 5

ASUS has launched a new product range featuring NVIDIA HGX H100 8-GPU AI servers, with pre-orders now extended into next year. Within this range, the ESC N8-E11 is a 7U server equipped with eight H100 Tensor Core GPUs, two 4th-generation Intel Xeon Scalable processors and a NIC topology, capable of meeting high-performance computing requirements.

ASUS’s NVIDIA-certified GPU servers offer full support for both Intel and AMD platforms, as well as PCIe interfaces. Furthermore, customers can opt for a direct-to-core (D2C) liquid cooling solution to significantly reduce the power usage effectiveness (PUE) of data centres.

The ESC8000 and ESC4000 are two NVIDIA-certified ASUS GPU servers that support both Intel and AMD platforms and provide PCIe interfaces. Among these, the ESC8000-E11 is a server supporting eight H100 PCIe GPU slots, featuring faster GPU interconnects and a higher-bandwidth architecture. Furthermore, the ESC8000-E11 can be configured with eight GPUs and an NVIDIA NVLink Bridge to meet the rapidly growing demands of AI and high-performance computing. Both small businesses and large-scale data centres stand to benefit from this.

ASUS stated that this new series of AI server products has attracted widespread attention, with pre-orders extending into next year. This initiative will further strengthen ASUS’s competitiveness in the AI sector and meet the ever-growing market demand.

DLC is an independent distributor specialising in electronic components, providing comprehensive supply chain solutions for OEMs, OBMs, EMS suppliers and design firms. We are committed to delivering high-quality products and accurate lead time planning at every stage of the electronic component distribution process, helping our customers minimise procurement risks in the spot market. Our core services include the supply of hard-to-source chips from spot markets, the recovery of slow-moving stock, procurement cost optimisation, and shared warehouse management. Through our global supply network and extensive market experience, we are able to provide rapid and accurate spot market intelligence. We also implement rigorous supplier audits and quality control inspection processes to ensure that every shipment meets quality standards, and we provide Certificates of Conformity (COC) to give our customers peace of mind when choosing DLC.