Right then,
Matt.
Here’s what matters today.
Your personal morning briefing, selected for you.
RIGHT THEN,
in 30 seconds.
AI-stock drops dragged global markets lower while oil jumped
Sharp falls in AI-related stocks overwhelmed global markets on 22 October as oil prices jumped on concern over the war with Iran. AP said technology weakness was strong enough to mask gains across many non-tech US stocks.
The sell-off followed a period in which AI enthusiasm had driven valuations and index concentration. Rising oil prices added a second shock by increasing inflation pressure and reducing the likelihood that central banks could look through price gains.
The episode shows how market concentration in AI and the oil shock can combine to hit portfolios, currencies and funding conditions at the same time.
Watch earnings from major AI-exposed technology firms and whether oil-price gains feed into inflation expectations.
US stocks fell as oil hit multi-year highs and AI volatility persisted
Wall Street closed broadly lower on 23 October as oil prices climbed to their highest level in more than four years, according to AP. The S&P 500 fell 0.2%, the Dow Jones Industrial Average dropped 541 points, or 1%, while the Nasdaq rose 0.2% as some large technology stocks steadied.
Brent crude traded around US$115 a barrel intraday after rising from below US$97 at the start of October. AP said three out of four S&P 500 stocks were lower, reflecting broad pressure from energy costs even as AI-related shares remained volatile after recent valuation concerns.
Higher oil prices and AI-stock swings are tightening global financial conditions, with spillovers to Asian currencies, inflation expectations and risk appetite.
Watch the coming US PCE inflation print and whether Brent holds above US$110 into the Fed meeting week.
US Commerce moved to award US$389 million for power-semiconductor and critical-material R&D
The US Commerce Department said on 22 October that it would award up to US$389 million in R&D funding for power semiconductors and critical materials. The proposed awards include up to US$73 million for Vertical Semiconductor to advance vertical gallium nitride power semiconductors for AI data-centre racks.
Power chips are becoming a bottleneck for AI infrastructure because data-centre racks require efficient conversion, thermal handling and high-voltage performance. Funding for gallium nitride and related technologies links the CHIPS agenda to grid resilience, fusion, advanced computing and defence applications.
The awards signal that the semiconductor race is shifting from leading-edge logic alone to power electronics that determine AI data-centre energy efficiency.
Watch award finalisation, technical milestones, private matching capital and whether Asian suppliers face new competition in power-device markets.
China’s draft AI export controls may raise costs for model developers going abroad
Analysts said on 17 October that China’s proposed AI export controls could impose short-term adaptation costs on domestic model developers targeting international markets. The draft covers 23 AI-related technologies including models, agents and chip designs, with possible thresholds tied to computing power, parameter count and global rankings.
The uncertainty is especially important for open-weight Chinese models deployed through global cloud providers, because exporting model weights could require licensing. A MOFCOM spokesperson said the rules were targeted and limited, and that most AI technologies would remain freely exportable.
Chinese open-weight models have become globally competitive, so licensing uncertainty could affect cloud marketplaces, enterprise AI adoption and US-China technology competition.
Watch how the final rules treat APIs, mobile apps, open-source releases and revenue-sharing arrangements with overseas hyperscalers.
China proposed export controls on 12 solar-manufacturing technologies
China proposed export controls on 12 technologies used to make solar equipment, according to a 18 October report. The proposal was part of a broader export-control revision covering AI, biotechnology and clean-energy manufacturing know-how.
Solar supply chains are already concentrated in China, especially in wafers, cells, equipment and process know-how. Controls on manufacturing technologies could slow attempts by other regions to localise solar production, even if finished panel trade continues.
The proposal affects energy-transition cost curves and industrial policy in ASEAN, Europe, India and the US, all of which want more domestic clean-tech capacity.
Watch the final Chinese control list, licensing exemptions, and whether non-Chinese solar equipment makers receive new government backing.
SK hynix’s Solidigm halted its IPO review after AI financing concerns hit chip stocks
SK hynix said on 23 October that Solidigm had halted its US IPO review after a sharp AI-financing rout hit semiconductor sentiment. The Philadelphia Semiconductor Index fell 6.7% on 22 October, while the Nasdaq dropped 2.6%.
Solidigm’s pause suggests that public-market windows for memory and AI-adjacent listings can close quickly when investors reassess data-centre economics. The development is particularly relevant because memory demand has been buoyed by AI servers, but valuations depend on confidence that capex will translate into cash flow.
A weaker IPO window can affect chip-sector funding, acquisition timing and investor appetite for AI supply-chain exposure across Asia.
Watch SK hynix commentary, memory pricing, other chip IPO filings and whether semiconductor indices stabilise after hyperscaler earnings.
US nuclear-reactor restarts advanced as AI data centres drove power demand
Three shut-down US nuclear reactors are being recommissioned to help meet AI data-centre electricity demand, according to a 20 October report. The projects together would bring about 2.25 GW of capacity back into service if completed.
The restarts show how AI load growth is reviving interest in firm low-carbon power, not just renewables and batteries. Recommissioning can be faster than new nuclear construction but still faces safety, licensing, financing and fuel-supply hurdles.
AI data-centre demand is changing electricity planning, and countries competing for data-centre investment will need credible firm-power strategies.
Watch Nuclear Regulatory Commission approvals, power-purchase agreements with hyperscalers and whether recommissioning costs stay within budget.
Ringgit opened firmer as AI worries weakened the US dollar
The ringgit opened higher on 23 October at 4.0785/4.0855 against the US dollar, compared with 4.0815/4.0850 at the previous close. Bank Muamalat’s chief economist linked the move to a softer greenback after equity prices fell on doubts about near-term returns from AI-related investment.
The same note said US two-year and 10-year Treasury yields had fallen by 16bp and 17bp to 4.64% and 4.95%, respectively, while the US Dollar Index eased to 101.625. The ringgit was mixed against ASEAN peers, gaining against the Thai baht and Indonesian rupiah but easing against the Singapore dollar.
Malaysia’s currency is being pulled into the global AI-capex and Fed-rate debate, not just domestic macro news, making US tech sentiment relevant to local funding conditions.
Watch the next Federal Reserve decision and US PCE inflation release for confirmation of whether rate-hike expectations continue to fade.
Johor data-centre boom could absorb most new Peninsular Malaysia power by 2031
Johor’s data-centre industry could require about 30TWh of electricity a year by 2031, equal to 18% of Peninsular Malaysia’s total supply, according to BMI research cited on 23 October. The state already has about 825MW of operational capacity, with another 3.9GW expected over the next three to five years as cloud and AI demand expands.
BMI warned that Johor faces tighter electricity-supply constraints than peer hubs including Sydney, Melbourne and Batam. It said the critical bottleneck is not headline generation capacity but the slower build-out of grid connections in concentrated areas such as Sedenak and Nusajaya, even as Tenaga Nasional plans RM43 billion of grid upgrades that can take five to 10 years.
Malaysia’s AI-infrastructure story is becoming an energy-infrastructure test: data-centre investment may be delayed, repriced or forced to fund grid upgrades if power access cannot keep pace.
Watch Tenaga Nasional’s grid capex phasing, connection approvals for Johor projects and the first 300MW tranche of YTL Power’s data-centre campus expected in 1Q 2027.
Singapore and the UAE agreed to launch economic-partnership negotiations
Singapore and the United Arab Emirates said on 21 October that they intend to launch negotiations on a Comprehensive Economic Partnership Agreement. Prime Minister Lawrence Wong and UAE President Sheikh Mohamed bin Zayed Al Nahyan also witnessed agreements on energy cooperation, water cooperation and disease-vector control in Abu Dhabi.
The proposed CEPA would target deeper trade and investment flows between a Gulf capital hub and Southeast Asia’s financial centre. Singapore said both sides see room to expand cooperation in AI, digitalisation, energy, defence and space, while also hoping to advance ASEAN-GCC free-trade talks when Singapore chairs ASEAN in 2027.
A Singapore-UAE CEPA would add another Gulf-ASEAN investment channel at a time when energy security, AI infrastructure funding and sovereign capital are strategically linked.
Watch for the formal negotiating mandate, tariff and services coverage, and whether ASEAN-GCC free-trade talks accelerate before Singapore’s 2027 ASEAN chairmanship.
Malaysia drafts data-centre law to manage energy, water and local value
Malaysia’s Economy Ministry said on 19 October that the government is drafting a Digital Infrastructure (Data Centre) Bill, expected to go to cabinet next year. The bill is intended to manage energy and water use, environmental impacts and the quality of investment in Malaysia’s digital ecosystem.
The ministry’s move followed Prime Minister Anwar Ibrahim’s call for closer scrutiny of infrastructure readiness, energy supply and resource efficiency. The government has also signalled that data-centre investors should pay for electricity-grid upgrades they require, and that power supply for the sector should not crowd out households or other industries.
The policy direction changes Malaysia’s data-centre proposition from low-cost expansion to regulated, resource-priced growth, with direct implications for hyperscalers, industrial parks and utilities.
Watch the cabinet submission timeline in 2027 and whether the bill sets binding power-usage effectiveness, water-usage or local-content requirements.
