Samsung Electronics has warned of weak sales for its AI chips in the first quarter, citing US export restrictions on high-bandwidth memory (HBM) chips to China. These restrictions, combined with the company’s struggle to meet Nvidia’s HBM chip requirements, are expected to impact demand. Samsung’s reliance on Chinese customers for about 20% of its HBM sales makes it particularly vulnerable. Despite this, the company plans to release upgraded HBM3E products in March to address shifting market demands.
The company also reported a 29% drop in operating profit for the fourth quarter, totalling 6.5 trillion won ($4.48 billion). Samsung expects limited earnings growth in the first quarter, driven by sluggish demand in the memory chip market, particularly for smartphones and personal computers. In addition, delays in GPU production, mainly due to strong demand and engineering challenges, have affected memory chip requirements for servers in data centres.
Samsung’s mobile phone division also faced challenges, with a 22% drop in operating profit in the fourth quarter. The company’s decision to use Qualcomm’s application processors instead of its own Exynos chips for the Galaxy S25 lineup has weakened its position in the mobile chip market. Despite these hurdles, Samsung anticipates a recovery in the memory market by the second quarter, depending on its ability to supply Nvidia with advanced HBM3E chips.
While competitors SK Hynix and TSMC have benefitted from the AI boom, with record profits, Samsung’s performance in the AI chip market has fallen behind. The company’s future performance will heavily depend on its ability to secure a stronger foothold in the rapidly growing AI sector.
India’s IT minister, Ashwini Vaishnaw, has praised the Chinese startup DeepSeek for its breakthrough in affordable AI, which has disrupted the sector with a powerful model costing just $5.5 million. He likened DeepSeek’s efficient approach to his government’s efforts to develop a local AI model through the IndiaAI mission, a $1.25 billion initiative to support AI startups and infrastructure development in India.
DeepSeek’s rapid success, claiming it took just two months to build its model using Nvidia’s H800 chips, has challenged the belief that China lags behind the US in AI. The startup’s app recently surpassed OpenAI’s ChatGPT in downloads on the Apple App Store, further highlighting its impact. Vaishnaw’s comments were seen as a response to remarks made by OpenAI’s CEO Sam Altman, who had previously expressed doubts about India’s ability to compete in the AI space with a $10 million budget.
Vaishnaw pointed out that while some questioned the government’s large AI investment, DeepSeek’s achievements prove that significant results can be achieved with more modest spending. As Altman prepares to revisit India in early February, his comments from last year continue to stir debate online, especially following DeepSeek’s unexpected success in the AI race.
Altman’s visit coincides with ongoing legal issues for OpenAI in India, as the company is engaged in a copyright dispute with local digital news and book publishers. The growing attention on AI developments in India underscores the shifting dynamics in the global AI race.
Top White House advisers have raised concerns over China’s DeepSeek using a technique known as “distillation” to potentially replicate US AI models, a method where one AI system learns from another. This could allow DeepSeek to benefit from the extensive investments made by US rivals, such as OpenAI, without incurring the same costs. DeepSeek recently made waves by releasing an AI model that rivals those of US giants, at a fraction of the cost, and giving away the code for free. US tech companies, including OpenAI, are now investigating whether DeepSeek’s model may have improperly used this distillation method.
Distillation, while common in the AI industry, may violate the terms of service of models like OpenAI’s. The technique allows a newer, smaller model to benefit from the learnings of a larger, more advanced one, often without detection, especially when using open-source models. Industry experts have pointed out that blocking such practices is difficult, particularly with freely available models like Meta’s Llama and French startup Mistral’s offerings. Some US tech executives, however, are advocating for stricter export controls and customer identification measures to limit such activities.
Despite the concerns, DeepSeek has not responded to the allegations, and OpenAI has stated it will work with the US government to protect its intellectual property. However, as AI technology continues to evolve, finding a way to prevent distillation may prove to be a complex challenge. The ongoing debate highlights the growing tensions between the US and China over the use of AI and other advanced technologies.
IBM’s fourth-quarter profits exceeded expectations, driven by a surge in software demand as businesses increased IT spending. Shares rose by about 10% in extended trading, reflecting investor confidence in the company’s strategy.
The software unit posted its strongest revenue growth in five years, as companies prioritised cloud infrastructure to support data-intensive generative AI applications.
Revenue growth of at least 5% is expected for 2025, outpacing the previous year’s 3% rise. IBM’s AI-related business, covering bookings and sales, has grown to over $5 billion, gaining $2 billion in just one quarter.
The company made its ‘Granite’ AI models open-source, contrasting with rivals that charge for access, positioning itself as a leader in open AI development.
Software performance contrasted with weaker consulting results, which declined by 2% to $5.2 billion. Around 80% of IBM’s AI business comes from consulting, with the rest from software.
While businesses focus on integrating AI into long-term projects, this shift has yet to translate into higher consulting revenue. Overall, total revenue remained steady at $17.55 billion, aligning with analyst estimates.
A global law enforcement operation has shut down a series of cybercrime websites used for selling stolen data, pirated software, and hacking tools. The FBI and Europol coordinated the takedown as part of ‘Operation Talent’, targeting platforms associated with Cracked, Nulled, StarkRDP, Sellix, and MySellix.
Seizure notices appeared on the affected websites, and officials confirmed that information on customers and victims had also been obtained. Europol stated that further details would be released within 24 hours, while the FBI has not yet commented on the operation.
Reports suggest that the targeted sites played various roles in the cybercrime ecosystem, facilitating the trade of stolen login credentials, compromised credit card details, and video game cheats. A message in a Cracked Telegram channel acknowledged the seizure, with administrators expressing uncertainty over the next steps.
Authorities continue to investigate, with the crackdown highlighting ongoing efforts to disrupt cybercriminal networks. More updates are expected as officials analyse the seized data and determine potential follow-up actions.
Shares of Nvidia rose in Europe on Wednesday, signalling a potential recovery after a sharp decline earlier in the week. The company, a key player in the AI sector, saw its Frankfurt-listed shares increase by 2%, following an 8.9% gain on Wall Street the previous day. This bounce came after a steep drop in Nvidia’s market value on Monday, triggered by the emergence of China’s DeepSeek AI tool, which posed a challenge to established players like OpenAI’s ChatGPT.
The decline in Nvidia’s stock earlier in the week saw the company lose nearly $600 billion in market value, marking the largest single-day loss in history for any company. However, markets showed signs of stabilising on Wednesday, bolstered by a surge in shares of ASML, the Dutch company that manufactures tools for chip production. ASML’s 11% jump helped lift European tech stocks, with chipmakers BE Semiconductor and ASM International also posting solid gains.
Investors seemed to regain confidence, with some believing that DeepSeek’s advancements might not disrupt the broader AI market. According to market strategist Chris Weston, the innovation from DeepSeek could even generate new demand for Nvidia’s GPUs, which are critical for AI applications. Meanwhile, Microsoft and OpenAI are investigating whether DeepSeek improperly used data from ChatGPT’s technology.
As markets remain volatile, investors are now looking ahead to earnings reports from major tech giants like Nvidia, Apple, and Microsoft, which could provide more clarity on the sector’s outlook. Despite ongoing uncertainties, the overall sentiment in the tech sector appeared more positive by midweek.
ASML, the leading maker of computer chip equipment, reported impressive fourth-quarter bookings of 7.09 billion euros ($7.39 billion), significantly surpassing analyst expectations. The surge in orders was driven by the increasing demand for advanced chip production tools fueled by the boom in AI. Following the announcement, ASML’s stock rose 11% to 722 euros, signalling investor optimism despite recent market volatility.
Analysts had expected bookings to total around 3.99 billion euros, up from 2.63 billion euros in the previous quarter. ASML’s CEO, Christophe Fouquet, emphasised that AI growth is a key driver for the industry, with major companies like TSMC, ASML’s largest customer, benefiting from the rise in AI chip demand. The company reported a net income of 2.7 billion euros on sales of 9.3 billion euros for the quarter, surpassing expectations.
Despite the positive results, analysts caution that the long-term impact of DeepSeek’s AI model—released this week and requiring less computing power than competitors—remains uncertain. Some fear it could affect the large investments tech giants make in AI chips. However, ASML’s strong quarterly performance and its 2025 sales forecast, ranging from 30-35 billion euros, suggest continued growth, particularly from companies like TSMC and SK Hynix.
ASML’s largest market was the US in the fourth quarter, accounting for 28% of sales, closely followed by China. However, the company anticipates that sales in China will decline, partly due to export restrictions on chip equipment imposed by the US and Dutch governments for national security reasons.
Alibaba launched a new version of its Qwen 2.5 AI model on Wednesday, claiming it outperforms competitors like DeepSeek-V3, GPT-4, and Llama-3.1-405B. The release, timed on the first day of the Lunar New Year when most Chinese workers are on holiday, highlights the growing pressure from DeepSeek’s rapid rise in the AI sector. The Chinese tech giant’s announcement emphasised that Qwen 2.5-Max delivers better performance across various AI benchmarks compared to some of the top models from OpenAI and Meta.
DeepSeek’s recent success, particularly after its January releases of the DeepSeek-V3 and R1 models, has shaken the AI market, including both international competitors and Chinese firms. The company’s low development and usage costs have raised concerns about the sustainability of large AI investments from US tech giants. The competition within China has intensified, with Alibaba’s Qwen 2.5-Max release and ByteDance’s update to its AI model shortly after DeepSeek’s R1 release, signalling a rapid response to the new market dynamics.
DeepSeek’s previous model, V2, had already disrupted the market last year, triggering a price war with Chinese firms slashing prices on AI models. Alibaba and other major tech companies, including Baidu and Tencent, had to follow suit, offering significantly cheaper options. Despite this, DeepSeek’s founder, Liang Wenfeng, has expressed that his company is focused on achieving Artificial General Intelligence (AGI) rather than competing on price, contrasting DeepSeek’s agile, research-driven approach with the more structured and costly operations of larger tech firms.
As the battle for AI supremacy intensifies, the emergence of DeepSeek, with its lean team of researchers, continues to challenge China’s tech giants, who may find themselves under pressure to innovate faster and more efficiently to keep up with the rapidly evolving AI landscape.
Coinbase has received regulatory approval to operate in Argentina, securing a Virtual Asset Service Provider licence from the National Securities Commission. The approval allows the exchange to offer its services within Argentina’s virtual asset framework, marking a significant step in its global expansion.
With inflation soaring and the local currency struggling, cryptocurrency adoption has surged in Argentina. Around 5 million Argentinians use crypto daily, viewing it as a tool to protect their wealth and access global financial markets. A recent Coinbase survey found that 87% of locals see digital assets as a way to enhance economic independence.
Coinbase’s operations in Argentina will be led by fintech expert Matías Alberti. The company plans to combine trading services with educational initiatives to help users navigate the risks and benefits of crypto. The company has highlighted security and compliance as key priorities, aiming to build trust in a market where financial stability remains a major concern.
Apple’s iPhone devices are now eligible to test SpaceX’s Starlink’s direct-to-cell service, which offers satellite-based coverage, according to T-Mobile. The trial, which started after receiving approval from the Federal Communications Commission (FCC) in November, currently provides “text via satellite” capabilities, with plans to add voice and data features in the future.
Initially, only select Android smartphones were eligible for the test, but T-Mobile has expanded compatibility to include iPhones running the latest iOS 18.3 software update. The partnership between Apple, SpaceX, and T-Mobile has been kept under wraps, with the companies working together to integrate Starlink support into Apple’s devices.
The FCC’s approval last year also allowed Starlink’s direct-to-cell service to assist in providing coverage in areas like North Carolina, which were severely impacted by Hurricane Helene. While Apple and SpaceX did not comment outside business hours, the initiative marks a significant step in expanding satellite coverage for mobile users.
