Outperformed the broader index after a remarkable recovery from Mar’s plunge. For strategic and security reasons, many heavy investments are earmarked to for frontend subsector which will likely lead to robust capex in near to mid-term. However, we do not see such catalyst for OSAT and view frontend’s forward integration as a threat and will erode its market share. While semiconductor sales forecast remains lacklustre, capital spending projection is encouraging reinforcing our position of favouring frontend over backend players. Maintain NEUTRAL on the sector with Frontken and UWC as our top picks.

Outperformed in 1H20. After a rapid recovery from Mar’s plunge, KLTEC gained 8% vs KLCI’s 6% decline (see Figure #1). Our earlier bearish view was proven right and share prices of our 3 SELL calls nosedived beyond our TP. Our only regret was not able to upgrade the sector timely enough when market sentiment turned around.

Blessing in disguise. Following the spark of US-China/Huawei tech war, many have rushed to develop their semiconductor capabilities, especially in leading edge (≤7nm) frontend fabrication (foundry) to be self-sufficient on the back of national strategic and security interests. As a result, this has led to huge investment commitments such as China IC Fund (USD29bn), American Foundries Act (USD20bn), CHIPS for America Act (USD10bn), TSMC’s Arizona foundry (USD12bn), SMIC’s Shanghai share sale (USD6.6bn), Taiwan subsidies (USD260m) and more. Coupled with the undersupply situation for advance node in the market, we opine that frontend subsector capex will be robust in the near to mid-term, benefiting companies such as Frontken, UWC and SAM Engineering (not rated).

How about OSAT? The solid frontend orders should trickle-down, benefiting backend players. However, heavy investment is not expected to be a catalyst or ploughed into this low-value segment which only contribute circa 10% of the value of an IC. As such, frontend’s gradual forward integration will eventually erode pure OSAT’s market share by providing one-stop and superior solutions (2.5/3D heterogeneous integration, HBM, chiplets and etc).

Global semiconductor sales. Despite achieving 6% growth to USD137bn in 4M20 (see Figure #2), the latest industry average growth projection is slashed from initial 7% to merely 2% (see Figure #3), this is even lower than our earlier conservative estimate of circa 5%. The growth is chiefly driven by memory (+15%) followed by logic and micro (+3%), offsetting the weaknesses in discrete (-7%), analog (-6%), opto (- 5%) and sensors (-2%) (see Figure #4). If memory prices continue to head south (see Figure #5), we do not discount another downward revision in global sales estimate. As for 2021, the sector is expected to pick up the slack in 2020 with an average growth of 11% (see Figure #6).

Equipment spending. Capital investment was strong in 5M20 with 21% gain in 3MA billings to USD12bn (see Figure #7). Based on SEMI’s latest forecast, there will be 13 new fab constructions (China×4, Taiwan×4, Americas×2, SEA×2 and Others×1) in 2020 with a total investment of USD38bn yielding additional installed capacity of 473kwpm (200mm equivalent). For 2021, 7 new fabs will be developed (Americas×2, China×2, Taiwan×1, Korea×1 and Others×1) with total capex of USD37bn with a total capacity of 289kwpm. This projection validates our view on frontend’s vigorous capex ahead and reinforces our bullishness on this sub segment.

Stronger greenback. HLIB expects USD to be stronger in 2020 averaging RM4.23- 4.28/USD compared to 2019’s average of RM4.14/USD (see Figure #8). As such, we expect tech firms to be marginally boosted thanks to their USD-denominated sales while partly offset by the USD cost items.

Source: Hong Leong Investment Bank Research - 13 Jul 2020