Right then,
Matt.
Here’s what matters today.
Your personal morning briefing, selected for you.
RIGHT THEN,
in 30 seconds.
China proposes export curbs on 40 technologies spanning AI, chips and solar
China said on Friday it plans to restrict or ban exports of 40 technologies covering artificial intelligence, semiconductors and solar power. The Ministry of Commerce and Ministry of Science and Technology published draft updates to China’s catalogue of technologies subject to export bans or restrictions.
The proposal includes 23 AI technologies, including advanced large models and agents, and would prohibit exports of core AI model technologies trained on sensitive national data. It also covers chip-design technologies and solar manufacturing know-how, including polysilicon-related processes.
The proposal marks a broader Chinese counterweight to US-led technology controls and could fragment access to AI models, chip design and clean-energy manufacturing know-how.
Watch the consultation process, final catalogue wording and whether named AI, chip or solar firms seek licences or restructure offshore technology transfers.
AI stocks swing after TSMC beat and CoreWeave sell-off
US markets steadied on Thursday despite sharp intraday moves in AI-linked stocks, with TSMC’s US-listed shares falling as much as 5.3% before closing down 1.8%. CoreWeave fell as much as 9.5% after a 15.5% tumble the previous day.
TSMC reported profit above expectations and said strong demand should continue in the final three months of 2026. Nvidia, the largest weight on the S&P 500, also declined, reflecting how concentrated AI exposure has become in major US indices.
The divergence between strong AI fundamentals and weak share-price reaction shows markets are now questioning valuation, leverage and funding assumptions around the AI build-out.
Watch CoreWeave funding conditions, Nvidia’s next earnings and whether AI infrastructure companies can refinance at acceptable spreads.
Zetrix faces RM2.16 billion sukuk obligations if covenant breach is not cured by October 22
Zetrix AI said it could face RM2.16 billion in sukuk payment obligations if it fails to resolve a covenant breach by October 22. The breach arose after forced share sales cut managing director Wong Thean Soon’s combined direct and indirect holding to 1.895%, below the requirement that he remain the company’s largest shareholder.
The company said the amount comprises RM2 billion of principal and RM164.5 million of future coupon payments. Zetrix shares fell 15.38% to 5.5 sen on Friday after touching five sen intraday, and the company is also under police investigation over unremitted Road Transport Department collections.
The case is a major test of governance, disclosure and creditor protection in Malaysia’s sukuk market, especially for digital-services issuers with public-sector collection roles.
The immediate trigger is October 22, when Zetrix must cure the breach or seek sukuk-holder approval for a waiver or further instructions.
Malaysia data-centre growth forecast at 33% CAGR despite reset from tighter rules
S&P Global Ratings projected Malaysia’s data-centre market to grow at a 33% compound annual rate from 2025 to 2030 on Friday, citing domestic demand and spillover from land- and utility-constrained Singapore. The rating agency also said the sector had entered a reset phase after rapid expansion strained power and water systems.
S&P said AI-related global data-centre demand is expected to grow strongly through 2030, with inference workloads forecast to expand faster than training and non-AI workloads. It also flagged rising off-balance-sheet lease obligations among hyperscalers, led by Microsoft, Meta, Oracle and Amazon, as a financing risk behind the infrastructure boom.
The forecast keeps Malaysia in the centre of ASEAN’s AI-infrastructure build-out, but it also underscores that grid, water and financing constraints are becoming binding limits rather than secondary issues.
Track Malaysia’s state and federal approvals for Johor, Cyberjaya and Kuala Lumpur data-centre projects, and whether new power or water rules change the economics of projects due before 2030.
Wall Street started the week near record highs before tech weakness took over
US stocks opened the week with a mixed finish on Monday while remaining near all-time highs. AP said roughly 64% of S&P 500 companies rose, led by communication-services stocks, even as investors awaited a busy week of earnings and macro signals.
The early-week resilience contrasts with the later technology-led sell-off, showing how quickly market leadership shifted from broad participation to AI and chip-sector concern. The sequence matters for assessing whether the pullback is a rotation or a broader de-risking.
Understanding the week’s progression helps investors separate a normal rotation near record highs from a more serious unwind of AI-linked valuations.
Watch breadth indicators, equal-weighted index performance and whether defensive sectors keep outperforming high-multiple technology names.
Meta’s open AI model release targets Chinese open-model competition
Meta unveiled a new open AI model to compete with Chinese rivals, with The Straits Times publishing the development on October 16. The release follows Meta’s own October 15 announcement that Muse Spark Open is available through multiple AI infrastructure and model-access platforms.
The competitive context matters because Chinese open models have gained developer usage and procurement attention. Meta’s release is aimed at keeping US open-weight models relevant for developers, enterprises and governments seeking alternatives to closed frontier systems.
Open-weight competition can influence AI sovereignty strategies, cloud spending and the balance between US and Chinese model ecosystems in Asia.
Watch download, API-usage and benchmark data for Muse Spark Open versus leading Chinese open models.
TSMC explores Texas expansion after record AI-driven profit
TSMC said it is exploring potential investments in Texas to expand US chip production after reporting record profit on strong AI demand. The company said July-September net profit rose 69% from a year earlier to NT$763.1 billion, while revenue rose 51% to NT$1.49 trillion.
TSMC said it is open to collaborating with Elon Musk’s planned Texas “Terafab” project but stressed that any TSMC plant must be run by TSMC. The company has already pledged US$265 billion for Arizona plants, including an additional US$100 billion commitment announced in July.
TSMC’s capacity decisions shape the geography of advanced-node supply for Nvidia, Apple and other AI customers, while also affecting Taiwan-US strategic dependencies.
Watch whether TSMC confirms a Texas site, its 2027 capital-expenditure guide and customer prepayment or long-term supply commitments.
TSMC reiterates US$60 billion to US$64 billion 2026 capex plan as agentic AI demand rises
TSMC reiterated on its October 15 earnings call that its full-year 2026 capital budget remains between US$60 billion and US$64 billion. Management linked the spending to robust multiyear demand from customers, including the emerging agentic AI market.
The company said higher capital expenditure is correlated with higher growth opportunities in following years, while also noting that leading-edge silicon demand is being driven by AI training and inference needs. The transcript gives investors a primary-source view of how TSMC frames AI as a structural rather than one-quarter demand driver.
The capex guide is a direct signal for semiconductor equipment makers, advanced packaging suppliers and ASEAN electronics exporters exposed to the AI cycle.
Watch TSMC’s January 2027 capex range, advanced-packaging bottlenecks and whether customers maintain order visibility into 2028.
Meta releases Muse Spark Open as a US open-weight frontier AI model
Meta released Muse Spark Open on October 15, saying the model is available through API platforms including Baseten, Fireworks AI, Groq, Modal, Nebius, OpenRouter and Together AI. Meta said it tested agentic workflows running on-device, including coding iOS changes and controlling DaVinci Resolve workflows.
Meta said the model was evaluated under its Superintelligence Scaling Framework and met its release thresholds. The launch is positioned as a US-developed open-weight alternative at a time when Chinese open models have gained global developer traction.
Open-weight frontier models affect enterprise procurement, sovereign AI strategies and the competitive balance between US and Chinese AI ecosystems.
Watch independent benchmark results, safety evaluations and enterprise adoption through the listed API providers.
Apple and Meta cooperate around AI despite long-running platform rivalry
Meta released Muse Spark Open on October 15, and Axios reported that the model is similar to Muse Spark 1.3 and is being positioned as the most powerful US-developed open-weight AI model available. The report framed the release in the context of Apple and Meta burying past tensions for AI-related collaboration.
The development highlights how AI model distribution and device integration are pushing former platform adversaries into pragmatic arrangements. It also reflects a broader industry race to place capable models closer to users and devices rather than only in cloud interfaces.
If major device and social-platform companies align around AI distribution, it could change app ecosystems, developer access and model-provider bargaining power.
Watch Apple developer announcements, Meta model integrations and whether on-device AI workflows become commercially available to users.
Indonesian stocks suffer worst one-day fall since 1998 after MSCI flags standalone downgrade
Indonesian equities plunged as much as 11.4% on Friday after MSCI said the market no longer qualified for emerging-market status and could be moved to standalone classification. The fall took the benchmark to its lowest level since early November 2020 and made it the worst one-day drop since January 1998.
MSCI’s move threatens capital outflows from passive and benchmark-tracking investors, and Jakarta stocks have now lost nearly 40% this year. The shock hit Southeast Asia’s largest economy at a time when regional investors are already reassessing liquidity, governance and index eligibility risks.
An MSCI reclassification would alter global fund allocation to Indonesia and could spill into ASEAN risk pricing, currencies and foreign-flow assumptions.
Watch MSCI’s formal consultation and implementation timetable, plus Bank Indonesia and Jakarta exchange measures to stabilise market liquidity.
