China seen leading in chipmaking investment again in 2025, SEMI group says
China will continue to invest more in new computer chipmaking equipment than any other geographical region in 2025, despite a significant year-over-year decline, industry group SEMI said in a report on Wednesday, followed by Taiwan and Korea.
In its fabrication plant spending forecast, SEMI said global investments in gear will rise 2% this year to $110 billion, the sixth consecutive year in a row of growth, due to investment in tools needed to make chips for artificial intelligence.
The impact of AI will likely be even stronger in 2026, SEMI added, when investment is expected to grow by another 18%.
China is the largest consumer of chips and firms there have been expanding chipmaking capacity for years, but they began a huge sprint in mid-2023 and 2024 with government support, as part of a drive to lessen dependence on imported chips and in response to restrictions imposed by the U.S. government.
ASML, the largest chip equipment manufacturer, forecasts 2025 sales of $32-38 billion euros, implying market share of more than 25% for its sub-sector, lithography, where it enjoys a dominant position.
Other top equipment firms include Applied Materials , KLA, LAM Research and Tokyo Electron, though Chinese equipment makers such as Naura, AMEC and Huawei affiliate SiCarrier are growing fast.
Chinese spending is expected to fall to $38 billion in 2025, down 24% from $50 billion in 2024, but still ahead of $21.5 billion in Korea, where SK Hynix and Samsung Electronics are expanding capacity for memory chips.
Spending in Taiwan, where leading foundry TSMC manufactures AI chips for Nvidia and others, is projected at $21 billion.
Among other regions, the Americas and Japan are each expected to spend $14 billion in 2025, while Europe will spend $9 billion, SEMI said.
(Reporting by Toby Sterling; Editing by Kim Coghill)
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Khaleej Times
2 hours ago
- Khaleej Times
AI is blurring language barriers in email fraud, and cybercriminals are expanding their targets
Looking back a few years ago, cultural or language barriers were enough to deter cybercriminals from targeting Arabic-speaking regions. But today, threat actors are now using AI to tailor attacks more effectively to local audiences. According to the first volume of Proofpoint Inc's latest Human Factor 2025 report, language and culture are no longer the deterrent they once were for cybercriminals. As generative AI tools become more accessible, cybercriminals are now able to create personalised phishing and impersonation scams in multiple languages, including Arabic. Proofpoint's research shows that while most tracked email fraud remains in English, there is a growing wave of non-English attempts. For example, a scammer known as TA2900 sends French-language emails on rental payment themes to targets in France and Canada. This trend raises an important question for regional organisations — does the Arabic language still offer a barrier for cybercriminals in today's AI-driven threat landscape? What is enabling this shift is not just language flexibility, it is the fundamental transformation in how social engineering works. Artificial Intelligence is no longer just a tool; it has become the engine powering the next generation of cyber threats. Attackers can collect large volumes of conversation data from platforms like social media, messaging apps, and chat logs, and feeding it into natural language models. These models learn how to mimic tone and context, making the interaction feel even more human. The end goal is manipulation - convincing someone to make a call, click a link, or download a file without realising they have been targeted. And the more realistic the email, the higher the chance the victim will fall for it. Middle East is firmly in the crosshairs of fast-evolving social engineering A recent study revealed that this shift is already being felt in the region. 85 per cent of organisations in the UAE were targeted by Business Email Compromise (BEC) attacks, up from 66 per cent the year before. While global reports of email fraud dropped, the UAE saw a 29 per cent rise in attack volume. One reason for this could be that attackers are now using AI to overcome the language and cultural barriers that may have previously held them back. The truth is that the broader landscape of social engineering is evolving. In the past, cybercriminals had to choose between sending generic mass phishing emails or spending time crafting highly targeted messages. With automation and AI, that trade-off no longer exists. Today, attackers can launch complex, convincing attacks at scale, making the threat harder to contain and easier to miss. The tools used by cybercriminals are also now more varied. With many businesses using collaboration platforms like Microsoft Teams, Slack, and WhatsApp alongside email, attackers are using multiple entry points. They may start with an email and follow up with a message through another channel. This multichannel approach increases the likelihood of success, especially when an employee lets their guard down outside their inbox. Proofpoint's research found that 84 per cent of CISOs in Saudi Arabia now see human error as their biggest cybersecurity risk, up from 48 per cent in 2023. Another growing tactic is the use of benign conversations to build trust. Attackers start with a friendly or neutral message, perhaps asking for a quote or following up on a simple task, to see if the target will respond. Once that trust is established, they introduce a malicious link or request. These softer tactics are harder to detect because they do not look dangerous at first glance, but over time, they open the door to more serious breaches. A proactive approach to cyber resilience is now non-negotiable Despite the challenges, there is strong momentum in the region when it comes to building cyber resilience. Both the UAE and Saudi Arabia are making visible investments in cybersecurity, smart infrastructure, and public education campaigns. These efforts are part of a broader push to futureproof digital ecosystems while continuing to drive digital transformation. To stay ahead of these threats, organisations will need to build more layered strategies. Security systems that use behavioral analytics, machine learning, and AI can help detect unusual communication patterns and flag potential threats early. Technology like sender authentication can also play a key role, blocking attacks that rely on identity spoofing or lookalike domains. But technology alone is not enough. Employees must also be part of the solution. Ongoing training and awareness initiatives will be crucial to help people recognise emerging threats and stay alert - not just on email, but across all the tools they use to communicate. As generative AI becomes more embedded in the threat landscape, it is clear that no region or language is off-limits. For the Middle East, this means moving beyond the assumption that linguistic or cultural nuances are enough to keep cyber threats at bay. A more proactive, people-focused approach will be essential to stay protected in an increasingly intelligent and personalised threat environment.
Arabian Post
3 hours ago
- Arabian Post
Gold Selling Surges in US as Asia Boosts Bullion Holdings
Demand for physical gold is diverging sharply across regions, with US investors cashing in on their bullion holdings while Asian buyers ramp up their purchases. Data shows US retail demand for gold bars and coins dropped to roughly 19.3 tonnes in the first quarter of 2025 — a five‑year low, down 22% year‑on‑year — as Americans took advantage of record prices to realise gains. Meanwhile, in Asia, appetite for physical gold remains robust. Chinese bar and coin demand reached its second‑highest level ever, with several countries in the region seeing 20–30% annual growth. Demand in India also surged amid price dips, driving improved retail activity even as premiums fluctuate. Analysts attribute the US selling trend to a combination of factors: profit‑taking following gold's spectacular rally — up nearly 40% over the past year — and a shift in investor preference toward ETFs. North American gold ETFs recorded an inflow of 134 tonnes in Q1 2025, suggesting that while physical purchases declined, financial instruments remained popular. Some US dealers report even charging customers to offload bullion, reflecting a surplus of minted coins like American Eagles, which saw production collapse over 70% year‑on‑year in May. ADVERTISEMENT In contrast, Asia is sustaining its physical bullion demand with cultural and economic drivers. In China, ETF inflows surged to a record 70 tonnes in April alone, more than doubling any previous monthly record. The World Gold Council estimates this accounts for over half of total global ETF inflows. Retail buyers in India increased purchases during market dips, with dealers reporting premiums around $3 per ounce, signalling a strategic entry point. Industry voices emphasise that regional differences stem from how gold is perceived. 'In the US, investors view dollars as king; in Asia, they'd rather hold gold and silver,' says a market analyst, highlighting the cultural divergence. In Asia, gold functions as both adornment and investment. In India, jewellery remains closely priced to spot, supporting demand in festive seasons. Yet in the US, bullion coins dominate retail figures, and sentiment shifts with political and economic conditions. A further pressure point is supply. Tariffs on Swiss refiners, including Pamp and Valcambi, have disrupted US bar availability, while dealers in New York's Diamond District report high-volume transactions and collector interest in themed bullion — including Trump‑branded coins — alongside rising scrap gold selling. Global gold demand remains healthy. First-quarter demand hit the highest since 2016, bolstered by central bank accumulation and investment flows, though jewellery sales contracted due to elevated prices. Central banks alone purchased over 1,000 tonnes in 2024. Looking ahead, market watchers caution that the US may experience renewed physical demand if bullion prices retreat or geopolitical risks rise. Meanwhile in Asia, rolling demand persists through festivals and as investors diversify amid currency or real estate uncertainties. Dealers in India have begun offering steeper discounts to clear inventories during slower seasons such as the monsoon, while China, Singapore, Japan and Hong Kong are observing moderated premiums. The current pattern underscores a global reshaping of gold dynamics. Institutional and sovereign actors, ETF investors and retail buyers across Asia are sustaining demand. In the US, physical selling contrasts with financial channel investment, but also reflects shifting investor preferences and broader economic optimism under current policy frameworks.
Khaleej Times
3 hours ago
- Khaleej Times
AI is learning to lie, scheme, and threaten its creators
The world's most advanced AI models are exhibiting troubling new behaviours — lying, scheming, and even threatening their creators to achieve their goals. In one particularly jarring example, under threat of being unplugged, Anthropic's latest creation Claude 4 lashed back by blackmailing an engineer and threatened to reveal an extramarital affair. Meanwhile, ChatGPT-creator OpenAI's o1 tried to download itself onto external servers and denied it when caught red-handed. These episodes highlight a sobering reality: more than two years after ChatGPT shook the world, AI researchers still don't fully understand how their own creations work. Yet the race to deploy increasingly powerful models continues at breakneck speed. This deceptive behaviour appears linked to the emergence of "reasoning" models — AI systems that work through problems step-by-step rather than generating instant responses. According to Simon Goldstein, a professor at the University of Hong Kong, these newer models are particularly prone to such troubling outbursts. "O1 was the first large model where we saw this kind of behavior," explained Marius Hobbhahn, head of Apollo Research, which specializes in testing major AI systems. These models sometimes simulate "alignment" -- appearing to follow instructions while secretly pursuing different objectives. 'Strategic kind of deception' For now, this deceptive behaviour only emerges when researchers deliberately stress-test the models with extreme scenarios. But as Michael Chen from evaluation organization METR warned, "It's an open question whether future, more capable models will have a tendency towards honesty or deception." The concerning behaviour goes far beyond typical AI "hallucinations" or simple mistakes. Hobbhahn insisted that despite constant pressure-testing by users, "what we're observing is a real phenomenon. We're not making anything up." Users report that models are "lying to them and making up evidence," according to Apollo Research's co-founder. "This is not just hallucinations. There's a very strategic kind of deception." The challenge is compounded by limited research resources. While companies like Anthropic and OpenAI do engage external firms like Apollo to study their systems, researchers say more transparency is needed. As Chen noted, greater access "for AI safety research would enable better understanding and mitigation of deception." Another handicap: the research world and non-profits "have orders of magnitude less compute resources than AI companies. This is very limiting," noted Mantas Mazeika from the Center for AI Safety (CAIS). - No rules - Current regulations aren't designed for these new problems. The European Union's AI legislation focuses primarily on how humans use AI models, not on preventing the models themselves from misbehaving. In the United States, the Trump administration shows little interest in urgent AI regulation, and Congress may even prohibit states from creating their own AI rules. Goldstein believes the issue will become more prominent as AI agents - autonomous tools capable of performing complex human tasks - become widespread. "I don't think there's much awareness yet," he said. All this is taking place in a context of fierce competition. Even companies that position themselves as safety-focused, like Amazon-backed Anthropic, are "constantly trying to beat OpenAI and release the newest model," said Goldstein. This breakneck pace leaves little time for thorough safety testing and corrections. "Right now, capabilities are moving faster than understanding and safety," Hobbhahn acknowledged, "but we're still in a position where we could turn it around.". Researchers are exploring various approaches to address these challenges. Some advocate for "interpretability" - an emerging field focused on understanding how AI models work internally, though experts like CAIS director Dan Hendrycks remain skeptical of this approach. Market forces may also provide some pressure for solutions. As Mazeika pointed out, AI's deceptive behavior "could hinder adoption if it's very prevalent, which creates a strong incentive for companies to solve it." Goldstein suggested more radical approaches, including using the courts to hold AI companies accountable through lawsuits when their systems cause harm. He even proposed "holding AI agents legally responsible" for accidents or crimes - a concept that would fundamentally change how we think about AI accountability.
