Yazılar

Samsung’s Q4 Profit Misses Expectations as Chip Issues and Rising Costs Weigh on Earnings

Samsung Electronics has reported a significant shortfall in its preliminary fourth-quarter operating profit, primarily due to challenges in its semiconductor business. The South Korean tech giant estimates an operating profit of 6.5 trillion won ($4.5 billion) for the three months ending Dec. 31, well below analyst expectations of 7.7 trillion won. Although the expected profit represents a 131% year-on-year increase, it marks a 29% decline compared to the previous quarter.

The company’s earnings were affected by rising research and development (R&D) costs and the ramp-up of manufacturing capacity for advanced semiconductors. Additionally, weak demand for conventional memory chips used in PCs and mobile phones further contributed to the dip in profits.

Samsung’s efforts to provide high-end chips to Nvidia have also posed challenges. Unlike its rival SK Hynix, which is Nvidia’s main supplier of high-bandwidth memory (HBM) chips used in AI GPUs, Samsung has struggled to meet the tech giant’s chip requirements. Nvidia’s CEO, Jensen Huang, mentioned that Samsung needs to “engineer a new design” to supply HBM chips, although he expressed confidence in Samsung’s ability to meet this challenge.

The disappointing earnings also extended to Samsung’s logic chip division, which designs and manufactures chips for mobile phones. Analysts estimate losses could have widened to about $1.5 billion in the fourth quarter due to lower production yields and reduced demand for mobile devices, including Samsung’s premium foldable phones.

Despite the weak earnings, Samsung’s shares saw a slight uptick, with analysts noting that the company’s woes were already factored into stock prices. Competition in the chip and mobile sectors remains intense, and analysts are cautiously optimistic that chip demand may have bottomed out.

 

Biden Signs Executive Order to Support AI Data Centers with Federal Power and Land

On Tuesday, President Joe Biden signed an executive order designed to bolster the infrastructure needed for advanced artificial intelligence (AI) data centers. The order, according to the White House, aims to address the growing energy demands of AI by leveraging federal land, particularly from the Departments of Defense and Energy, to host gigawatt-scale AI data centers and new clean power facilities.

Biden emphasized that the initiative would accelerate the development of AI infrastructure in the U.S., promoting economic competitiveness, national security, AI safety, and clean energy. “The next generation of AI infrastructure will be built here in America,” Biden stated, underscoring the importance of aligning the country’s energy and technological sectors.

A key provision of the order mandates that companies using federal land for AI data centers must purchase a portion of American-made semiconductors. The exact number of chips required will be determined on a case-by-case basis for each project. This comes as part of the Biden administration’s broader push to invest over $30 billion in subsidizing U.S. chip production.

Tarun Chhabra, White House technology adviser, pointed out that the increasing demand for computational power to train and operate advanced AI models necessitates the creation of robust infrastructure. By 2028, leading AI developers will require data centers with up to five gigawatts of capacity to support the most sophisticated models.

The executive order also addresses national security concerns by ensuring AI technology remains within the U.S. and allied nations, as the Commerce Department moves forward with additional restrictions on AI chip and technology exports. Chhabra noted that AI systems already present substantial risks, including potential military applications and threats related to biological, chemical, radiological, or nuclear weapons.

The order also instructs agencies to expedite electric grid interconnection, permitting processes, and transmission development surrounding the newly designated federal sites.

Semiconductor Groups Criticize Biden Administration’s New Export Controls

A coalition of semiconductor and manufacturing trade groups has expressed concern over new export control regulations announced by the Biden administration, claiming they were rushed through without proper industry consultation. In a private letter dated January 13 and addressed to U.S. President Joe Biden, groups such as the Semiconductor Industry Association (SIA) and SEMI voiced objections to new rules that could affect the global distribution of advanced computing chips.

The Biden administration’s regulations, released on Monday, introduced a three-tier system governing the placement of U.S.-made chips, including those from companies like Nvidia, in AI data centers. These rules mandate that most countries will need to obtain licenses to use these chips in such projects.

In their letter, the trade groups expressed concerns over an additional rule expected to be announced soon, which would impose even stricter controls on high-bandwidth memory, a critical component for AI chips. The groups argued that these regulations were being implemented without adequate input from industry stakeholders, raising fears about the impact on U.S. companies and the potential to shift market share to international competitors.

High-bandwidth memory is primarily produced by U.S. and South Korean firms, and the new rules could limit its sale to China. A source familiar with the upcoming regulations indicated that the measures might also reverse a previous interpretation, potentially reducing revenue for companies like Lam Research, which supplies chip-making equipment to China. Lam Research has yet to comment.