News related to
chip

The facility, launched on August 18, expands the company’s Korea region only about 14 months after its second data centre began operating. Alibaba Cloud established its first South Korean data centre in 2022 after entering the market in 2016, reflecting a faster infrastructure build-out as demand for generative AI and enterprise AI agents increases.

The third site adds capacity for computing, storage, networking, databases, security, containers and cloud-native services. It will also support workloads requiring lower latency and stronger disaster-recovery arrangements, while giving customers more options to distribute applications across separate facilities.

Alibaba Cloud has not disclosed the new centre’s power capacity, investment cost, number of graphics processing units or current utilisation rate. The company has indicated that additional computing equipment can be installed as customer demand expands.

Operating three data centres also changes the resilience of Alibaba Cloud’s South Korean infrastructure. With two facilities, disruption at one centre could leave the remaining location handling workloads alone. The addition of a third allows applications to remain distributed between two facilities if another suffers an outage.

The expanded configuration has enabled Alibaba Cloud to raise the service-level availability commitment for its Korea region from 99.95% to 99.99%, strengthening its pitch to companies running business-critical applications and AI services.

Enterprise AI is becoming an increasingly important part of that strategy. Alibaba Cloud is introducing six agentic AI services in South Korea covering development, testing, deployment, operations and cybersecurity. They include AgentRun, STAROps, ACS Agent Sandbox, Agent Security Center, AI Security Guardrails 2.0 and Agentic SOC.

The services are aimed at companies seeking to move AI agents from experimental projects into production environments, where governance, reliability and security become more demanding. Enterprises are increasingly looking beyond access to large language models towards infrastructure capable of operating autonomous or semi-autonomous software agents at scale.

Alibaba Cloud is pairing those services with its broader full-stack AI offering, including infrastructure, development platforms and the Qwen family of AI models. Qwen models have recorded more than 3 billion downloads globally, while developers have created more than 300,000 derivative models based on the family.

South Korean companies are already using parts of Alibaba Cloud’s AI ecosystem. Gwangju-based Gendive has integrated Alibaba models including Wan for video generation and HappyHorse for image generation into its DevDive enterprise automation platform. NAVER Z’s ZEPETO platform also uses Alibaba Cloud infrastructure to support real-time 3D avatar generation and immersive services across Asian markets.

Alibaba Cloud has built local partnerships with companies including Megazone Soft and Team Sparta, seeking to combine its infrastructure with consulting, implementation and industry-specific services. Such alliances are important in a South Korean cloud market where global hyperscalers compete with strong domestic technology groups.

Data sovereignty is another factor shaping cloud investment. Businesses handling sensitive information increasingly want workloads hosted within national borders, particularly as regulators pay greater attention to cybersecurity, resilience and governance surrounding AI systems. Alibaba Cloud says its South Korean infrastructure is designed to comply with applicable local data-protection and cybersecurity requirements.

The Korea expansion forms part of Alibaba’s wider commitment to spend about 380 billion yuan, roughly $53 billion when the programme was announced, on AI and cloud infrastructure over three years. The group has already committed substantial capital to computing capacity, chips and AI development as it attempts to establish cloud services as a larger growth engine.

Alibaba Cloud revenue rose sharply in the April-June quarter of 2026, with cloud and AI-related sales reaching about 48.4 billion yuan and growing 45% from a year earlier. The expansion has come at a cost: Alibaba’s quarterly capital expenditure climbed to nearly 67.7 billion yuan as spending on computing infrastructure and semiconductor components increased.

The company has also expanded cloud infrastructure in France, Japan, Malaysia and Mexico. With the new South Korean facility, Alibaba Cloud says its global network has reached 104 availability zones across 30 regions.

The Malaysian Palm Oil Board has developed Sawit EcoTherm, a non-conductive dielectric fluid made largely from palm-derived materials that allows servers and other computing components to be immersed directly in coolant. The technology eliminates the need for conventional evaporative cooling towers and is being positioned for AI and hyperscale facilities where rising processor densities are producing substantially more heat.

Sawit EcoTherm remains at the pre-commercial stage, with pilot projects, technical validation and partnerships with data centre operators among the steps required before widespread deployment. Researchers say the fluid could reduce cooling-related energy consumption by about 30% to 50%, largely by removing server fans and transferring heat directly from electronic components.

MPOB estimates that conventional water-based cooling across Malaysia’s expanding data centre industry could eventually require as much as 876 million litres of water a day. Other projections based on currently identified developments are lower, illustrating the uncertainty surrounding future capacity and cooling configurations. Data centres planned across Selangor, Johor and Negeri Sembilan have been estimated to require more than 445 million litres daily if projected demand materialises.

The pressure is particularly acute in Johor, which has become one of Asia’s fastest-growing data centre clusters because of its proximity to Singapore, available industrial land and established power and connectivity infrastructure. Water availability and grid capacity have increasingly become constraints as developers seek to commission large AI facilities.

Malaysia’s data centre and cloud computing investment approvals reached RM144.4 billion between 2021 and the middle of 2025. Global technology and infrastructure groups including Microsoft, Amazon, Google, ByteDance, Equinix and AirTrunk have committed billions of dollars to expanding computing capacity in the country.

AirTrunk announced plans this year to invest about $3 billion in two additional hyperscale campuses in Johor, adding more than 280 megawatts of capacity. Equinix separately unveiled an investment exceeding $190 million for another Kuala Lumpur facility designed to support more than 2,200 cabinets and equipped for advanced liquid cooling.

Those developments are accelerating interest in alternatives to traditional air and water cooling. AI servers can generate several times more heat than conventional computing equipment, making direct liquid cooling and immersion systems increasingly attractive as rack power densities rise.

Sawit EcoTherm uses palm-derived ester formulations engineered to provide electrical insulation while absorbing heat from processors, memory, power supplies and graphics processing units. MPOB has developed two formulations, EcoTherm 5s and EcoTherm 8s, both with dielectric breakdown strength exceeding 75 kilovolts.

The fluids are designed for single-phase immersion systems, where coolant remains liquid while circulating through a closed loop. The 8s formulation has a minimum flash point of 200 degrees Celsius, while the 5s version has a minimum of 160 degrees. Their characteristics are intended to combine thermal performance with relatively low pumping requirements.

Palm-derived material can also offer a cost advantage. Commercial synthetic and petroleum-based immersion fluids can cost about RM50 to RM80 per kilogram, while the raw-material cost for producing the palm formulation has been estimated at roughly RM20 to RM25 per kilogram. Final commercial pricing has not been announced.

MPOB says up to 95% of the material used in Sawit EcoTherm can come from renewable bio-based sources. The formulations are biodegradable and free from PFAS, a class of persistent chemicals that has attracted growing regulatory scrutiny because of its resistance to environmental degradation.

The technology nevertheless carries broader sustainability questions because expansion of oil-palm cultivation has historically been associated with deforestation, biodiversity loss and pressure on wildlife habitats. The environmental case for palm-based coolant will therefore depend partly on traceable sourcing, lifecycle emissions and whether demand can be met without encouraging additional land conversion.

Immersion cooling itself is not new. Vegetable-derived, mineral-oil and synthetic dielectric fluids are already available, while many operators developing AI facilities favour direct-to-chip liquid cooling because it can be integrated into server architectures without fully submerging equipment. Commercial success for Sawit EcoTherm will depend on compatibility with hardware, long-term fluid stability, maintenance requirements and acceptance by server manufacturers.

Under the MOU, VinFast and Gowa Motor Group will establish a joint venture to develop VinFast’s dealership network across Indonesia, with a target of opening at least 30 showrooms and service centers.

The partnership reflects the two companies’ shared commitment to building a long-term strategic relationship in Indonesia’s growing electric vehicle market. By partnering with VinFast, Gowa Motor Group demonstrates its confidence in VinFast’s vision, capabilities, and the long-term potential of its electric mobility ecosystem.

The new network will serve as a key pillar of VinFast’s long-term growth strategy in Indonesia, offering VinFast’s electric vehicles alongside aftersales services, maintenance, and customer care that meet the company’s global standards. The partnership will contribute to VinFast’s broader plan to expand its dealership network in Indonesia. The company currently operates more than 40 showrooms nationwide and aims to add over 150 new showrooms in the coming years.

The agreement with Gowa Motor Group reinforces VinFast’s strategy of partnering with leading local companies to expand its distribution network, improve customer access to high-quality products and services, and support Indonesia’s transition toward greener transportation.

Gowa Motor Group is an established automotive group in Indonesia with extensive experience across multiple automotive brands and vehicle segments. Its operation spans automotive distribution, dealership operations, passenger vehicles, and commercial vehicles, supported by a broad network and strong understanding of the local market. With its automotive expertise, operational capabilities, and market knowledge, Gowa Motor Group is expected to become a strategic partner for VinFast in expanding its distribution network, enhancing the customer ownership experience, and accelerating the adoption of electric vehicles in Indonesia.

Mr. Antonio Zara, Chief Executive Officer of VinFast Southeast Asia, said: “Gowa Motor Group is an established automotive group in Indonesia with extensive experience across multiple automotive brands. We believe the combination of VinFast’s product development capabilities, advanced technologies, and comprehensive EV ecosystem with Gowa Motor Group’s local market expertise, automotive network, and operational experience will provide a strong foundation for expanding VinFast’s presence in Indonesia. More importantly, this partnership will enable us to deliver an increasingly convenient, comprehensive, and high-quality EV ownership experience for Indonesian customers.“

Mr. Sadikin Aksa, Commissioner of Gowa Motor, said: “We see tremendous potential in Indonesia’s electric vehicle market and strongly believe in VinFast’s long-term vision. Our partnership not only creates new business opportunities for both companies but also reflects our shared commitment to accelerating Indonesia’s green transition. Leveraging our experience across multiple automotive brands and our established network in Indonesia, Gowa Motor Group looks forward to working alongside VinFast to bring sustainable mobility solutions and high-quality services closer to consumers across the country.“

Indonesia is one of the key pillars of VinFast’s global growth strategy. Since entering the market, VinFast has rapidly expanded its product portfolio, commenced operations at its manufacturing facility in Subang, and built a comprehensive EV ecosystem through a growing network of strategic partners, including charging infrastructure developer V-Green, all-electric ride-hailing service GSM, financial institutions, banking partners, authorized service workshop partners, and dealerships nationwide. By combining manufacturing capabilities, a comprehensive EV ecosystem, and an expanding partner network, VinFast is steadily advancing toward its goal of becoming one of the leading electric vehicle brands in Indonesia and across Southeast Asia.

Hashtag: #VinFast

The issuer is solely responsible for the content of this announcement.

About VinFast

VinFast (NASDAQ: VFS), a subsidiary of Vingroup JSC, one of Vietnam’s largest conglomerates, is a pure-play electric vehicle manufacturer with the mission of making EVs accessible to everyone. VinFast’s product lineup today includes a wide range of electric SUVs, e-scooters, e-bikes, and electric buses.

In Indonesia, VinFast is expanding its electric mobility ecosystem through a growing dealership and after-sales network, local partnerships, and its manufacturing facility in Subang, West Java, reinforcing its long-term commitment to the market.

Learn more at: https://vinfastauto.id

About Gowa Motor Group

Gowa Motor Group is an established automotive group in Indonesia with extensive experience in the automotive sector. The Group represents and operates across multiple automotive brands and vehicle segments, including passenger and commercial vehicles, supported by dealership, distribution, and customer service capabilities. With a broad network across Indonesia and a strong understanding of the local automotive market, Gowa Motor Group continues to build long-term partnerships and expand access to quality mobility products and services.

Eleven shares changed hands at 1.168 million won each at 8 am Seoul time on August 6, about 30% below the previous Korea Exchange closing price. The stock quickly recovered most of the loss and ended Nextrade’s 50-minute pre-market session down roughly 2%, indicating that the extreme opening price was not driven by a change in the company’s business outlook.

The episode echoed a July 28 disruption when a single SK Hynix share traded at 1.272 million won, nearly 30% below the preceding close. That transaction triggered a volatility interruption on Nextrade and transmitted the distorted price into cryptocurrency derivatives linked to the chipmaker.

A perpetual futures contract tracking SK Hynix fell about 19% during the first event. Almost $60 million of leveraged long positions were liquidated within minutes, leaving more than 900 traders with estimated realised losses of $17.4 million. The platform operating the contract later offered to compensate eligible customers and said it would reassess the way external prices were incorporated into its valuation mechanism.

Thursday’s plunge again occurred at the start of Nextrade’s pre-market window, when order books are typically shallower than during the main session. The exchange allows stocks to move as much as 30% above or below the previous Korea Exchange closing price, meaning a small order can reach the limit when few competing bids are available.

The trade underscored the distinction between a price that is technically valid and one that accurately reflects the market’s consensus. Electronic systems can treat an executed transaction as authoritative even when only a handful of shares are involved. Prices may then spread through data feeds, trading algorithms, structured products and blockchain-based markets before human dealers can assess whether the move is credible.

Nextrade launched in March 2025 as the first serious challenger to the Korea Exchange’s long-standing dominance. It attracted investors with lower fees and trading hours stretching from 8 am to 8 pm, compared with the shorter core session on the main exchange. Retail traders account for most activity on the platform, which has at times handled a substantial share of the country’s equity turnover.

Its expanded trading day has improved access and competition, but the two SK Hynix incidents highlight risks when extended hours are combined with fragmented liquidity. Order books can be especially vulnerable at the opening instant, when market makers have not fully populated bids and offers and overseas price signals may still be feeding into domestic systems.

Regulators and exchange operators now face pressure to consider safeguards based on trade size, available depth and deviation from a volume-weighted reference price. Measures could include auction-style openings, minimum order-value requirements, stronger market-maker obligations or delayed dissemination of extreme prices produced by very small transactions.

The disruption comes during an unusually volatile period for SK Hynix and South Korean technology shares. The chipmaker has been at the centre of investor enthusiasm over high-bandwidth memory used in artificial-intelligence accelerators, but its shares have also recorded repeated double-digit moves as traders reassess valuations and future demand.

SK Hynix reported a sharp year-on-year increase in quarterly profit, supported by high-bandwidth memory sales, yet the results fell short of elevated market expectations. The disappointment contributed to a broad sell-off that erased trillions of dollars from South Korean equities at its peak and pushed measures of expected volatility to exceptional levels.

Leveraged exchange-traded funds linked to SK Hynix and Samsung Electronics have amplified those movements. Authorities have halted new sales of some single-stock leveraged products and proposed tighter investment limits, larger minimum deposits and higher transaction costs after retail investors suffered heavy losses.

Research into the market turbulence has also raised concerns that predictable rebalancing by leveraged funds can encourage traders to buy or sell ahead of compulsory end-of-day transactions. Such feedback loops may magnify price changes even when there is little new information about a company’s operations.

Zhongji Innolight dropped about 10% in both Shanghai and Hong Kong, while Eoptolink Technology also lost 10%. Suzhou TFC Optical Communication declined roughly 6%. The CSI 300 Telecommunication Services Index fell as much as 9% during early trading before trimming some losses.

The sell-off followed disclosures that the administration of US President Donald Trump is drafting a measure that would bar imports of new Chinese optical transceiver models on national security grounds. The Federal Communications Commission is working on the proposed restrictions and is seeking to publish them before the end of the year.

No final decision has been announced, and the proposed rules could be altered, delayed or abandoned. Neither the White House nor the FCC has publicly detailed the measure.

Optical transceivers are essential components in modern data centres. They convert electrical signals into light and allow vast volumes of information to move at high speed through fibre-optic cables connecting servers, switches and computing clusters.

Demand for the devices has surged as technology groups build larger facilities for training and operating artificial intelligence models. These systems require thousands of advanced processors to communicate rapidly, making optical networking a critical part of the AI supply chain.

Chinese manufacturers have secured a substantial position in the global market by combining high-volume production with competitive pricing. Zhongji Innolight holds an estimated 27% share of the global data-centre transceiver market and is among the leading suppliers of high-speed products used in AI computing systems.

The company derived 62% of its revenue from the United States during the first quarter of 2026. It has previously warned that an escalation in trade tensions could produce a steep decline in earnings or lead to losses.

Eoptolink is similarly exposed, generating about 96% of its sales from overseas markets. TFC Optical also depends heavily on foreign customers, leaving all three companies vulnerable to any restriction affecting shipments to US cloud-computing operators.

The proposed action would mark a further expansion of Washington’s technology controls. Earlier measures concentrated mainly on preventing China from acquiring advanced semiconductors, chipmaking equipment and other technologies with potential military applications.

US policy is now extending towards limiting Chinese companies’ access to the domestic market. The FCC has already imposed or prepared restrictions covering telecommunications equipment, drones, routers, connected power inverters and advanced robots.

Under the plan being considered, the agency could initially prohibit imports of all new optical transceiver models and then grant exemptions to suppliers outside China. Existing devices may remain unaffected, although the scope of any final regulation has yet to be established.

US officials are concerned that foreign-made components embedded in data centres could enable data theft, malware installation or service disruption. Supporters of tighter controls argue that Washington should act before Chinese equipment becomes deeply integrated into critical AI infrastructure.

The approach reflects lessons drawn from Huawei, whose telecommunications equipment became widely installed before security restrictions were introduced. Removing or replacing those systems proved costly and complicated for network operators.

A ban would also create challenges for US cloud-computing companies. Amazon Web Services and other operators may have to shift orders to suppliers such as Coherent and Lumentum, potentially increasing procurement costs and slowing the expansion of data-centre capacity.

Shares of US optical-networking companies rose after details of the proposal emerged. Lumentum gained about 7%, Coherent climbed 11% and Applied Optoelectronics advanced 18%, as investors anticipated greater demand for non-Chinese products.

Industry specialists have cautioned that US suppliers may lack the manufacturing scale required to replace Chinese vendors quickly. Any abrupt change could tighten global supplies of high-speed transceivers at a time when technology groups are spending heavily on AI infrastructure.

Some market analysts also viewed Wednesday’s declines as excessive. They said the proposal could serve as a negotiating tool before an expected visit by Chinese President Xi Jinping to the United States in September.

Trade relations remain shaped by mutual dependence. Washington relies on China for rare-earth minerals used in electronics and optical systems, while Chinese technology companies depend on US customers, processors and specialised equipment.

The most severe damage was reported from Idukki, Kottayam and Pathanamthitta, where saturated hillsides collapsed, rivers crossed warning levels and rescue teams moved families from vulnerable areas. The state government directed district administrations to prepare for emergencies and assigned ministers to coordinate relief operations in badly affected areas.

Sumathi, a resident of Kudayathoor near Thodupuzha, was killed when a landslide struck her house at Adurmala around 2 am. Her husband Ravi and son Ratheesh were pulled from the debris with injuries. The building was destroyed, and rescuers recovered Sumathi’s body after a search lasting several hours.

Another landslide at Kolahalamedu in Vagamon killed 72-year-old Prabhakaran Nair. A mass of earth crashed through a window and buried the room where he was sleeping. Fire and Rescue Services personnel, police and residents used an earthmover to clear the wreckage before recovering his body.

Joseph Johnny and his mother Regina Johnny were also killed after a landslide buried a house near Poonjar in Kottayam district. Search and rescue teams were deployed to the area as heavy rain continued to restrict access through hilly roads.

The weather department issued red warnings for Thrissur, Malappuram, Kozhikode, Wayanad, Kannur and Kasaragod, where isolated rainfall exceeding 20 centimetres within 24 hours was considered possible. Pathanamthitta, Kottayam and Idukki were also placed under red warnings for rainfall ranging between 12 and 20 centimetres.

Orange warnings covering Thiruvananthapuram, Kollam, Alappuzha, Ernakulam and Palakkad indicated the likelihood of very heavy rainfall. The warnings covered nearly the entire state and reflected the possibility of flash floods, landslides, uprooted trees, power disruption and damage to weak structures.

An offshore trough extending from south Gujarat to Kerala, combined with strong lower-level westerly winds, intensified the rainfall. Rain or thundershowers were forecast across most parts of Kerala through the weekend, although the heaviest phase was expected on Saturday.

Flooding spread across central districts as rivers and streams overflowed. Erattupetta was inundated after the Meenachil River breached its banks, while the Mundakkayam causeway went under water. Sections of the Pala-Poonjar road became impassable, affecting movement between several towns and villages.

Ranni in Pathanamthitta faced extensive flooding, with water entering the Ittiyappara bus stand, shops and houses. Landslides were reported from surrounding highland areas, while the Pamba and other rivers rose rapidly following overnight rainfall.

Thodupuzha town and nearby settlements experienced waterlogging after streams spilled into residential colonies. The Thodupuzha River crossed the danger mark, while the Muvattupuzha River moved above its warning level. Residents along riverbanks and in low-lying areas were told to prepare for evacuation.

Authorities increased controlled releases from the Malankara Dam as inflows surged. Four shutters that had been opened by 1.5 metres were scheduled to be raised in stages to 2.5 metres. Officials warned that the release could further increase levels in the Thodupuzha and Muvattupuzha rivers.

Road travel was disrupted across Idukki after landslips blocked the Moolamattom-Vagamon, Vagamon-Ettumanoor, Vagamon-Upputhara and Kattappana-Vazhavara routes. Debris also halted traffic near Machiplavu on the Kochi-Dhanushkodi National Highway, where a car and a heavy vehicle were caught in a landslide.

The benchmark closed at 61,434.19, down 930.73 points, or 1.49%, after briefly dropping to 60,448.90. The index moved through a range of almost 2,700 points, having reached a session high of 63,138.04 during early trading.

The intraday slide from the high exceeded 1,700 points, reflecting unusually volatile trading rather than a fall of that size from Tuesday’s closing level. The index had ended the previous session at 62,364.92 after losing more than 2,500 points.

Semiconductor companies led the decline as investors questioned whether the vast sums being invested in AI infrastructure would generate returns quickly enough to justify elevated technology valuations. The selling was amplified by weakness in South Korea and Taiwan, whose markets contain some of the world’s largest memory-chip and contract-manufacturing companies.

Asian chipmakers had benefited from expectations that spending on data centres, advanced processors and high-bandwidth memory would continue expanding. Those assumptions are now facing closer scrutiny as capital expenditure rises, borrowing costs remain high and earnings expectations become more difficult to exceed.

The shift was visible in South Korea, where the Kospi suffered another severe fall and chip heavyweights came under sustained pressure. Taiwan’s benchmark also declined as investors sold shares exposed to the global semiconductor supply chain.

Tokyo-listed equipment makers and chip-related companies were caught in the same retreat. Their large influence on the price-weighted Nikkei magnified the index’s movements, leaving the broader market vulnerable to concentrated selling in a small group of expensive stocks.

Investors were also awaiting quarterly results from major US technology companies, including businesses whose spending plans have become central to demand forecasts for advanced chips. Any indication that data-centre expenditure is slowing could deepen concerns about the durability of the AI investment cycle.

The Nikkei’s decline followed a powerful rally that had carried the benchmark above 70,000 earlier in July. Despite the pullback, the index remained substantially higher than a year earlier, showing how quickly valuations had expanded during the AI-driven advance.

Currency movements added another layer of uncertainty. The yen remained near its weakest level against the dollar in about four decades, despite gaining slightly during Wednesday’s trading. A weaker currency generally supports exporters by increasing the yen value of overseas earnings, but an extreme decline raises import costs and complicates monetary policy.

Japan depends heavily on imported energy and raw materials. The yen’s depreciation, combined with higher oil prices, threatens to intensify inflation and squeeze household spending. It has also increased speculation that authorities may intervene in foreign-exchange markets.

Markets are watching for any disruption to the yen carry trade, in which investors borrow cheaply in Japan and direct the funds towards assets offering higher returns elsewhere. A sudden strengthening of the yen can force traders to unwind those positions, producing rapid declines across equities, bonds and currencies.

The Bank of Japan begins a two-day policy meeting on Thursday. Traders are assessing whether persistent currency weakness and rising energy costs could push officials towards tighter policy, even as instability in equity markets increases the risks surrounding another rate move.

Oil prices climbed more than 3% as conflict in the Middle East threatened shipping and energy supplies. Higher crude prices added to inflation concerns ahead of the US Federal Reserve’s policy announcement and reduced investors’ willingness to hold high-valuation growth stocks.

The dollar remained supported near a one-month high as traders considered whether the Federal Reserve could maintain a restrictive policy stance. Higher US yields tend to pressure technology valuations by reducing the present value of expected future earnings.

The Monetary Authority of Singapore said uncertainty over whether companies can sustain vast spending on data centres, advanced chips and computing infrastructure has become a significant risk to the economic outlook. The warning reflects growing scrutiny of whether revenue generated by AI services can justify the scale of capital committed by technology groups.

MAS Managing Director Chia Der Jiun said the global economy should continue to benefit from AI-related expenditure in the near term, but the longer-term impact remains difficult to assess. A sudden reduction in investment could affect manufacturing, trade and financial asset prices, particularly in economies closely linked to electronics supply chains.

The assessment carries particular significance for Singapore, where technology-related production has driven much of the economic expansion this year. The economy grew by about 6 per cent during the first half of 2026, while AI-linked electronics accounted for more than 70 per cent of export growth. Semiconductor production, data-centre equipment and related services have helped offset pressure from trade restrictions, higher energy costs and geopolitical instability.

MAS expects economic activity to remain firm during the second half of the year, supported by technology, construction and financial services. However, it cautioned that tighter global financing conditions or a reversal in AI expenditure could undermine the durability of that growth and ease inflation more rapidly than expected.

The warning comes as the world’s largest technology companies raise spending on processors, servers, power systems and data centres. Capital expenditure by major AI-linked companies is projected to rise by more than 75 per cent to around $700 billion in 2026. Large technology groups are also turning increasingly to debt markets to finance the buildout, strengthening the connection between AI investment, credit conditions and the wider economy.

Fitch Ratings has identified an AI market correction as an emerging global credit risk, citing elevated valuations, concentrated investment and uncertainty over future returns. Amazon and Alphabet are among companies whose combined bond issuance has reached about $182 billion, underlining the growing role of borrowed funds in supporting infrastructure spending.

Supporters of the investment surge argue that generative AI could raise productivity, reduce costs and create new markets comparable to earlier technological transformations. Demand for advanced chips and cloud capacity remains strong, while business investment data continue to show momentum. US orders for core capital goods rose 0.9 per cent in June, and shipments increased 1.9 per cent, their strongest gain in four and a half years, helped by spending on computers and electronic equipment.

Questions persist over how quickly that spending will generate sustainable profits. Data centres require large quantities of electricity, specialised chips and cooling infrastructure, while rapid improvements in computing equipment can shorten replacement cycles. Companies may face pressure to keep investing even before earlier projects achieve their expected returns.

Academic research indicates that the AI boom combines strong technological foundations with pockets of speculative vulnerability. Enterprise adoption, revenue growth and productivity improvements support parts of the valuation expansion, but investment has outpaced monetisation in some segments. Private-market valuations are also concentrated among a limited number of companies, increasing the risk that a reassessment of growth expectations could spread rapidly.

Global equity markets are already showing greater sensitivity to concerns about returns from AI spending. Semiconductor shares have faced selling pressure, while technology indices have retreated from earlier peaks. Investors are closely examining earnings from leading cloud and chip companies for evidence that demand remains strong enough to support infrastructure budgets.

Singapore is vulnerable to such shifts because of its position in global electronics, finance and trade networks. A decline in chip orders could affect factory output and exports, while a fall in technology valuations could weaken investment sentiment and household wealth. Financial institutions could also face losses through corporate lending, bond holdings and market exposure.

MAS has separately highlighted cybersecurity threats arising from AI and quantum-computing advances. AI tools can improve fraud detection and network monitoring, but they can also make phishing, impersonation and hacking attempts more convincing. The authority has established an industry task force to help financial institutions strengthen their defences as technology-driven risks become more sophisticated.