· We attended Day 2 of AiSEM@Selangor 2026 event, and we gathered several themes that support our sector call.
· Malaysia holds 13% of global ATP but captures only 6% of value add, so policy now favours design and equipment. Government is now backing IC design and advanced packaging to move Malaysia up the value curve.
· WSTS estimates an approximately US$2.1tn chip market in 2027, but component lead times are stretching to 7-18 months.
· We maintain OVERWEIGHT on the Malaysian technology sector and keep all ratings and target prices unchanged. Given both VITROX and MI have hit our TP, we will reassess ratings and target prices on the upcoming quarterly results.
Session takeaways from AiSEM@Selangor
Selangor bets on front-end IC design (YB Ng Sze Han, Selangor EXCO)
Ng positioned Selangor as Malaysia’s front-end IC design hub, complementing Penang’s strength in manufacturing and ATP. Malaysia holds about 13% of global ATP but captures only about 6% of industry value, versus 57% for chip design and IDMs. Selangor is committing RM50m to a new RM150m Strategic Investment Fund alongside Penang and Khazanah.
· IC Design Parks 1 and 2 now host 20 companies and 409 engineers, ahead of the 400 year-end target.
· ASEM has trained 850 engineers, with 100% job placement and starting pay above RM5,000.
From "Made in Malaysia" to "Made by Malaysia" (YB Liew Chin Tong, Deputy Minister of Finance)
YB Liew Chin Tong framed Budget 2027 as an effort to link the domestic capital market to the chip sector and reduce reliance on MNC capital. He announced a National Technology Financing Centre to build tech-lending capability in banks. He positioned Malaysia as a neutral, trusted node between the US and China. He argued Malaysia has a pay problem rather than a talent problem, pointing to 1.1m Malaysians working in Singapore.
· Policy direction is to finance local chip champions, not just host MNC capacity.
· A RM2,500 graduate pay floor for firms with turnover above RM50m was cited; details are still pending.
Powering the Era of AI (Dr Jeehoon Kang, FuriosaAI)
Kang said LLM token demand rose about 20x in a year as agentic AI workloads run in continuous loops. About 70% of the 100GW+ of data-centre capacity expected by 2030 will serve inference. FuriosaAI’s RNGD chip (TSMC 5nm, 180W) entered mass production in 2026. Kang claimed 1.6-2.1x more tokens than NVIDIA’s RTX 6000 Pro at equal latency, with LG, Samsung SDS and Upstage as customers.
· Inference efficiency (tokens per watt) is becoming the main competitive metric.
· Next-generation chips are being built with Broadcom and TSMC; Kang said the company is in talks with MIMOS and SIDEC.
Virtual Twins for Semiconductors (Samuel Tan, Dassault Systemes)
Tan argued that Malaysia can move up the value chain faster by simulating fabs, equipment and process recipes before building them. He cited existing users including Lam Research (5,500+ users), KLA and Samsung. He claimed customers have cut packaging costs by 20-40% and process costs by 15-30%.
· Digital twins cut costly physical prototyping and could speed up training toward the NSS target of 60,000 engineers.
The Eyes and the Brain of AI (Dr Ren Wu, Lucitrum)
Wu said physical AI is constrained by sensing, not computing, because standard RGB cameras miss information that multispectral sensing captures. Lucitrum is building multispectral sensor chips in Malaysia to serve robotics. Wu sized sensing at 10-30% of an estimated US$25tn robotics market by 2050.
· Malaysia could become the supplier of “eyes” for global physical AI, building on its existing supply chain.
The Race for Semiconductor Resilience (moderated by McKinsey)
Speakers from Intel, X-FAB, Sandisk, QES and GreatAsic agreed that resilience means being hard to replace. That comes from engineering ownership and local IP, not capacity alone. About 90% of the audience polled believe Malaysia can be a critical node, mainly by moving up the value chain. Panellists split on whether a domestic leading-edge fab is needed: QES argued yes, GreatAsic argued design-led companies can thrive without one.
· Supply tightness is real: QES said precision-robot lead times have stretched to 7-9 months, and GreatAsic said substrates need orders placed 12-18 months ahead.
· QES has sold locally designed fab equipment to European and US customers, but cited talent retention against Singapore pay as a constraint.
From Funding to Scale (moderated by SIDEC)
Cradle, the Securities Commission, CREST and FuriosaAI said capital is available. The gap is entrepreneurial and governance capability among engineer-founders. Cradle said 36 of its top 100 startups earn more than RM1m in net profit, and about 80% of its prototype grants commercialise. A MITI co-investment fund now matches private semiconductor raises 2:1.
· Funding programmes are now in place, including MyChipStart, the BIG programme and Co-Invest; execution is the bottleneck.
· Panellists named policy continuity across elections as the top ask from global investors.
The Rise of Autonomous Machines (moderated by Cyberview)
eMooVit, Alphaswift, ALS, TCab Tech and Grab showed physical AI moving from pilots into early operations. Examples include autonomous public transport in Putrajaya, a drone blood delivery in Pahang and robovan pilots. Panellists said the main blockers are sandboxes with no regulatory exit path, manual approvals for every route, and thin VC patience.
· Public trust, built on safety and cybersecurity, is the precondition for scaling.
· The demand read-through for sensors, edge chips and power semiconductors is real but long-dated.
Disclaimer
The report is for internal and private circulation only and shall not be reproduced either in part or otherwise without the prior written consent of Apex Securities Berhad. The opinions and information contained herein are based on available data believed to be reliable. It is not to be construed as an offer, invitation or solicitation to buy or sell the securities covered by this report.
Opinions, estimates and projections in this report constitute the current judgment of the author. They do not necessarily reflect the opinion of Apex Securities Berhad and are subject to change without notice. Apex Securities Berhad has no obligation to update, modify or amend this report or to otherwise notify a reader thereof in the event that any matter stated herein, or any opinion, projection, forecast or estimate set forth herein, changes or subsequently becomes inaccurate.
Apex Securities Berhad does not warrant the accuracy of anything stated herein in any manner whatsoever and no reliance upon such statement by anyone shall give rise to any claim whatsoever against Apex Securities Berhad. Apex Securities Berhad may from time to time have an interest in the company mentioned by this report. This report may not be reproduced, copied or circulated without the prior written approval of Apex Securities Berhad.
| Currency | Buy Rates (RM) | Sell Rates (RM) |
|---|---|---|
| USD | 4.065852 | 4.098389 |
| EUR | 4.573697 | 4.582846 |
| CNY | 0.609357 | 0.610574 |
| HKD | 0.518023 | 0.522177 |
| SGD | 3.170371 | 3.195870 |