• The AI semiconductor supercycle remains firmly intact, supported by record AI infrastructure spending. WSTS now forecasts global semiconductor sales to surge 89.9% YoY to US$1.5tn in 2026F, while TSMC and hyperscalers continue to expand capex aggressively. Malaysia is already benefiting, with E&E exports rising 39.5% YoY in January 2026. 

• Despite the KLTEC Index trading above its historical average valuation, we believe the premium is justified as earnings have yet to fully reflect the current AI-driven upcycle. We would turn more cautious only if AI capex weakens materially, the ringgit appreciates sharply, or WSTS significantly lowers its industry outlook. 

• Maintain OVERWEIGHT on the Malaysian technology sector for 2H2026.

• We raise our ViTrox TP to RM9.35 (from RM7.04) and maintain BUY, reflecting stronger earnings visibility from AI-driven advanced packaging demand. We also upgrade Inari Amertron to HOLD (TP: RM2.11) on improving RF and photonics prospects, and QES Group to BUY (TP: RM0.74) following stronger order momentum and higher FY27F earnings forecasts.

WSTS Raises Outlook                          

WSTS sharply raised its 2026 global semiconductor sales forecast to US$1.5tn (+89.9% YoY), driven primarily by an unprecedented recovery in the memory market. Memory is expected to more than triple on AI-related HBM and DRAM demand, while logic and microprocessors also remain key growth drivers.

AI Capex Boom                                      

The semiconductor upcycle is being underpinned by a surge in hyperscaler AI investment, with US CSP capex projected to rise 71% YoY in 2026 and the four largest hyperscalers expected to spend around US$725bn in capex. Capacity constraints, rather than weak demand, remain the key bottleneck, supporting strong multi-year demand for semiconductor equipment, as echoed by ViTrox and Mi Technovation's management.

Packaging Bottleneck                         

AI chip production remains constrained by advanced packaging capacity, with TSMC's CoWoS facilities fully booked through 2026 despite aggressive expansion. This supply bottleneck aligns with ViTrox's order backlog and longer procurement lead times, reinforcing strong demand for semiconductor inspection and packaging equipment.

Malaysia Well Positioned                  

Malaysia is well positioned to benefit through advanced packaging, AI infrastructure, semiconductor equipment demand and ongoing supply-chain diversification. This is already evident in the country's strong E&E export growth (+39.5% YoY in January 2026), rising semiconductor investments under the National Semiconductor Strategy, and favourable exposure across our coverage, including Frontken, Mi Technovation, ViTrox, EG Industries, Inari, QES Group and Wentel.

Premium, But Justified

We maintain our OVERWEIGHT stance on the technology sector as the KLTEC Index's 29.5x FY27F forward P/E remains supported by the structural AI-driven semiconductor upcycle, improving earnings momentum and multi-year earnings visibility.

MI and EG as our top picks: Our highest-conviction BUYs remain Mi Technovation and EG Industries. ViTrox moves to our highest conviction BUY alongside MI and EG following this revision.

Top-Down Thesis: The AI/Semiconductor Supercycle in Numbers

WSTS has sharply upgraded its 2026F global semiconductor sales growth forecast to 89.9% YoY (from 26.3% YoY previously), with the market now expected to reach US$1.5tr in 2026F (vs. US$795.6bn in 2025) before a further 27% YoY expansion to US$1.91tr in 2027F. The upgrade is overwhelmingly memory-led: memory segment sales are forecast to surge 249% YoY to US$803.9bn in 2026F (vs. US$230.0bn in 2025), reflecting both AI-driven HBM/DRAM demand and a tightening pricing cycle. Logic (+37% YoY) and Micro (+20% YoY) segments are also inflecting meaningfully, while Discrete, Sensors and Optoelectronics grow at a more modest high single-digit pace.

This memory-and-logic-led upcycle is underpinned by continued hyperscaler capex: US cloud service provider (CSP) capex is estimated to grow 71% in 2026F (vs. 61% in 2025), with 1Q26 capex already up 80% YoY. ViTrox management corroborated this directly in its 27 April briefing, citing CSP capex exceeding US$500bn for AI servers as underpinning multi-year demand visibility for its Automated Board Inspection (ABI) and Machine Vision System (MVST) franchises.

We see four transmission channels through which this supercycle flows into our Malaysian coverage: (i) advanced-node and advanced-packaging equipment/materials demand (Frontken, Mi Technovation, ViTrox); (ii) AI data centre and optical networking infrastructure (EG Industries, Inari's photonics segment); (iii) memory and advanced packaging supply tightness pulling forward equipment capex (QES Group, Wentel); and (iv) geographic supply-chain diversification toward Malaysia and Thailand (EG Industries' Prachinburi facility, ViTrox's planned 100%-owned China entity, Mi Technovation's Johor plant).

The advanced-packaging component of that budget is funding a near-quadrupling of TSMC's CoWoS capacity — from c.35,000 wafers/month in late 2024 to a targeted 130,000 wafers/month by end-2026. Management has publicly acknowledged that CoWoS capacity remains sold out through 2026 despite this expansion, which is precisely the dynamic ViTrox's own management described in its most recent briefing (RM300m of 1Q26 purchase orders that could not be fully fulfilled; procurement lead times extended from 6 to 9 months). We read this as direct, bottom-up corroboration of the top-down capex data: the bottleneck in AI chip production today sits in advanced packaging and the inspection/test equipment that qualifies it, which is exactly the layer of the value chain where Frontken, ViTrox and Mi Technovation sit.

Demand for this capacity is itself underwritten by an extraordinary step-up in hyperscaler capital spending. The four largest US hyperscalers (Amazon, Microsoft, Alphabet and Meta) are guiding to a combined c.US$725bn of capital expenditure in 2026, up c.77% from c.US$410bn in 2025 and c.US$226bn in 2024 — a more than 3x increase in two years. Several of these companies have explicitly cited capacity constraints (rather than demand uncertainty) as the binding limit on their build-out pace, which is consistent with the supply-constrained commentary we are hearing directly from our own covered companies (ViTrox, Mi Technovation) rather than being a generic industry narrative.

Where Malaysia sits in this build-out

Malaysia's own trade data corroborates that this capex is already flowing through into the local supply chain rather than remaining a forward-looking thesis. Malaysia's total exports grew 19.6% YoY in January 2026, beating market expectations of 13.7% and marking the strongest annual export growth since September 2022, with electrical & electronics (E&E) exports — the segment encompassing the bulk of our coverage — the standout contributor at +39.5% YoY. Integrated circuits alone made up roughly a third of total exports in the period. Malaysia currently handles an estimated 13% of global semiconductor assembly, testing and packaging (ATP) activity, and the government's National Semiconductor Strategy (launched 2024) has to date attracted in excess of RM63bn in approved semiconductor-related investment, with a stated ambition to double Malaysia's share of the global semiconductor market from c.7% to c.14% by 2029 and to reach RM1tn in E&E exports by 2030.

Premium, But Justified

The KLTEC Index is currently trading at 29.5x FY27F forward P/E, modestly above its five-year historical average but still below the +1 standard deviation valuation band. While sector valuations are no longer inexpensive, we believe the premium is justified by the structural AI-driven semiconductor upcycle and the sector's improving earnings outlook. WSTS has upgraded its 2026 global semiconductor sales growth forecast to 89.9% YoY, while Malaysia's E&E exports have already accelerated 39.5% YoY as of January 2026, signalling that earnings momentum is beginning to flow through the domestic technology supply chain.

Although the index appears optically expensive on a 61x trailing P/E—61% above its five-year historical average and above the +2 standard deviation band—we place greater emphasis on forward earnings, as trailing earnings have yet to capture the sharp earnings acceleration currently underway. As such, we believe the forward P/E provides a more representative measure of sector valuation at this stage of the semiconductor cycle.

More importantly, earnings visibility is unusually strong by historical standards, which supports a structurally higher valuation multiple. During a typical semiconductor cycle, OSAT companies generally provide one to two quarters of earnings visibility, while equipment manufacturers carry order backlogs of only two to three quarters. The current cycle is materially different. ViTrox's ABI segment has secured a RM233m backlog with customer visibility extending into 2027, Frontken is benefiting from the accelerated expansion of TSMC's Plant 2, while EG Industries has secured a confirmed USD241.6m purchase order with a 12-month delivery schedule. This level of company-backed, multi-year earnings visibility is uncommon in previous semiconductor cycles and, in our view, supports sector valuations remaining above historical averages.

We maintain our OVERWEIGHT stance on the technology sector. However, following the sector's strong re-rating, we believe investors should become increasingly selective, as the remaining upside is likely to be concentrated in companies with clear earnings catalysts, robust order visibility and proven execution, rather than being broadly available across the sector.

Valuation Framework: P/E Band Discipline Across Coverage

Our underlying principle remains that a re-rating toward +2SD should be reserved for names where (i) the earnings base is inflecting structurally, (ii) there is a clear, dated catalyst the market can underwrite, and (iii) the current multiple is not already pricing in that optionality. This update revises that assessment for three names — ViTrox, QES and Inari — based on new evidence; our view on the remaining six names is unchanged.

ViTrox

We maintain BUY on ViTrox with a higher TP of RM9.35 (from RM 7.04), based on the 5-year historical mean PE of 54.67x applied to a blended FY27F/FY28F EPS of 17.1 sen (FY27F: 15.95 sen; FY28F: 18.31 sen). Using a blended FY27F/FY28F EPS better captures ViTrox's actual earnings trajectory than a single-year snapshot. The company is in a structural earnings ramp (FY25A: RM133.2m core NP → FY26F: RM237.1m → FY27F: RM289.4m implied on revised 15.95 sen EPS), and the 2026–2028 capex cycle visibility (TSMC commitment through at least 2027, ABI backlog with customers booked into 2027) justifies forward blending. Anchoring to FY27F alone understates the earnings power that will be visible to the market by mid-2027.

ATECH

ATECH’s 1QFY26 core net profit missed on USD/MYR weakness and supply chain disruption (gold, memory shortages), but we view this as transitory, with no material order rescheduling. The new automotive segment (TPMS, IoT auto-modules) and debt-free balance sheet (RM200.1m cash) provide diversification and optionality. We maintain BUY with an unchanged TP of RM0.87 on an unchanged PE 18.1x multiple applied to rolled-forward FY27F EPS.

EG

EG Industries delivered a record 3QFY26 (PATAMI +75.7% YoY) on a fundamentally shifting product mix toward higher-margin optical modules. On 14 May 2026, EG secured a confirmed USD241.6m (~RM950m) purchase order for 800G optical modules with a 12-month delivery timeline — the single largest, most directly quantifiable AI-infrastructure capex catalyst across our entire coverage. We maintain BUY with an unchanged TP of RM2.69, based on an unchanged 14.5x FY27F EPS — still the most conservative multiple in our coverage relative to the magnitude of the earnings re-rating, and the clearest remaining case for valuation (not just earnings) re-rating over 2H26.

FRONTKN

FRONTKN's AGTC TFT/LCD cleaning line relocation and Plant 2 expansion were both brought forward from the original 2027 timeline, a signal of robust forward demand visibility ahead of the 2nm production ramp into 2H26; a new Singapore semiconductor customer began contributing revenue from May 2026, ahead of original guidance. We maintain BUY with an unchanged TP RM5.71, based on 45x FY27F core EPS of 12.7 sen — Frontken's 5-year historical mean.

INARI

We upgrade INARI from SELL to HOLD and raise our TP to RM2.11, based on a higher 37.7x P/E FY27F core EPS of 5.6 sen (forecast unchanged pending formal re-underwriting), where the revised multiple sits at approximately Inari's own 5-year historical mean PE (c.38x). We view a move to the historical mean — rather than a re-rating beyond it — as the appropriately calibrated response to genuinely improved, management-confirmed visibility, while we have not yet raised our FY27F core net profit estimate of RM213.2m; we prefer to do so only once we see RF utilisation tracking toward the 70–75% guided range in early FY27F print(s), at which point we would look to move toward a full BUY.

MITECH

MITECH delivered a record 1QFY26 (core net profit +61% YoY) despite Q1 being seasonally weakest, with SMBU PBT already running at ~35% of our full-year segment forecast. SEBU's Mi Series WLCSP die sorters are directly levered to advanced packaging; SMBU's Accurus Taiwan facility is running at maximum capacity serving a Taiwan foundry's expansion; the proposed SGX listing of SMBU is a sum-of-parts re-rating catalyst not reflected in our single-entity PE valuation. TP RM6.23, based on 37x FY27F EPS — already at +2SD of the 3-year mean, so further re-rating should come from earnings delivery and further announcement on its SGX listing proposal rather than multiple expansion. We maintain BUY with an unchanged TP RM6.23, based on 37x FY27F core EPS of 16.8 sen.

WENTEL

WENTEL’s 1QFY26 margin compression (GP margin to 24.1%) was disappointing, but management guides recovery from 2QFY26 on bigger-scale E&E machine orders; precision engineering peers UMS and Coraza have reported strong results, a positive read-across. We maintain BUY with an unchanged TP RM0.39 on an unchanged 20x FY27F PE.

QES

We upgrade QES from HOLD to BUY with a higher TP of RM 0.74 (previous TP RM 0.54) based on a higher FY27F EPS of 3.68sen (previously FY2F EPS 2.7sen), unchanged 20x FY27F sits between -1SD (17.72x) and -0.5SD (21.61x) on a forward basis — the most conservative possible BUY target within the band, well below mean. we remain positive on QES's medium-term prospects underpinned by the recovering semiconductor orderbook, and improving Glenmarie utilisation.

Key Risks to Our Sector View

• MYR appreciation against USD/TWD: OSAT players are typically the most sensitive to forex movements given USD-denominated revenue against MYR-denominated cost bases; EMS players are comparatively insulated given cost pass-through mechanisms to customers.

• Semiconductor capex cycle slowdown or memory price normalisation: our top-down thesis leans heavily on the WSTS 249% YoY 2026F memory growth forecast holding; any moderation would disproportionately hit the highest-multiple names (ViTrox, Mi Technovation) first.

• Execution risk on the specific catalysts underpinning this update's two rating changes: ViTrox's China entity establishment and humanoid robotics ramp are both early-stage; Inari's RF circuit-content win is contingent on the September 2026 flagship launch proceeding as currently guided by its key customer.

• Supply chain and component disruption: ViTrox itself flags shortages across manpower and materials (including Intel-alternative chips, memory, GPUs, CPUs, cameras), forcing build-ahead inventory and extended procurement lead times; Aurelius Technologies and Wentel have both already been directly affected this cycle.

• Geopolitical and tariff risk: US-China trade tensions, potential US semiconductor tariffs, and Middle East-related logistics disruption all pose risk to the supply-chain re-routing thesis underpinning Malaysia's positioning, including ViTrox's planned China entity.

• Execution risk on the Inari RF circuit-content guidance: our upgrade rests on management's own guidance of a 1→4 circuit increase for the September 2026 flagship; any slippage in the launch timeline, a weaker-than-guided ramp, or a shift back toward lower circuit content would directly undermine the case for holding the multiple at the historical mean rather than below it.