Semiconductor 2H26: Look Before You Leap

Executive Summary

We reiterate the Semiconductor proxy VanEck Semiconductor ETF (SMH) at Neutral, with a target price at USD 717.

Our message is straightforward: while our target price implies upside, we maintain a Neutral rating based on valuation discipline.

We nonetheless hold 2.5 stars ‘Neutral’ and do not recommend overweighting at current valuations.            

YTD market performance

2026 has been an exceptional year. The industry is up ~90% year-to-date, reached an all-time high of USD 671.83 on 22 June, and even after a ~9% pullback it sits among the best-performing corners of global equities.

Beneath the surface, the story is dispersion and rotation. The index return conceals an enormous spread between constituents. Memory was the defining trade of the year; the equipment complex and the former laggards re-rated hard; and, strikingly, the prior leader Nvidia was close to flat.

Table 1: SMH constituent YTD movement

Source: Bloomberg Finance, Data as of 30 June 2026.
*KLA Corp experienced a 10-for-1 stock split in May hence the figure is not a reflection of its YTD performance.

What has changed since our last update

The marathon continues, but current valuations warrant a more measured pace.

Our last published view was the 2Q26 outlook, “Semiconductor 2Q26: Entering into the Second-Half Marathon” (April 2026)".

It’s central framing that the AI-driven up-cycle is a marathon, not a sprint: durable demand, but a longer, more measured and selective back half that rewards endurance over chasing gives us a clean yardstick.

The second-half-marathon call has aged well: the cycle proved durable and structurally improved, and the choppiness we cautioned on has duly appeared. The two things that evolved are that leadership broadened (Nvidia lagged, memory and equipment led) and that valuation has caught up.

Despite the market rally beyond our initial expectations, our bullish thesis has proven correct, and our positioning - neutral on fabless while overweighting foundry and memory, has largely been validated.

Read more: Semiconductor 2Q26: Entering into the second half marathon

Figure 1: SMH 1y performance and our previous call

Investment Thesis

Three reinforcing reasons we like the exposure, even as the price keeps us disciplined.

First, memory has inflected and is becoming structurally less cyclical. Micron’s Q3 blowout (revenue USD 41.5 bil, EPS $25.11, ~83% cloud margin) reads as a regime signal, not a one-off: HBM is booked into 2028 with ~40% of revenue under floor-priced multi-year contracts (~USD 100 bil RPO), shifting the most profitable memory from a spot commodity toward a contracted component.

That softening of cyclicality underpins our decision to apply a full 24x fair PE on FY28 rather than haircut it, and its read-through lifts pricing, deposition intensity (Lam) and EUV demand (ASML memory is now 51% of system sales) across the basket.

Figure 2: Memory Pricing Trend

Figure 3: Memory AI-exposure trends

Figure 4: AI-related storage demand forecast

Second, an AI-capex supercycle underwrites every layer of the fund. Hyperscaler spending above USD 725 bil for 2026 and WFE above USD 140 bil is the common denominator beneath GPUs, custom silicon, foundry, memory and equipment — and crucially it is largely contracted, visible in TSMC’s CoWoS sold out through 2027, ASML’s ~€38.8bn backlog, Broadcom’s >USD 100 bil FY27 line-of-sight and AMD’s multi-gigawatt OpenAI and Meta commitments. The nuance we respect is that this spend is concentrated in a handful of buyers, so any one of them signalling digestion would pressure the whole complex.

Figure 5: Hyperscaler CAPEX forecast

Third, June’s reconstitution improved the mix, lifting Applied Materials, KLA and ASML to the front and the WFE trio to ~15% combined. Equipment is the highest-quality layer of the buildout, an oligopoly (ASML’s EUV monopoly; near-duopolies in deposition, etch and process control) with pricing power, high incremental margins and a fresh advanced-packaging growth leg and it captures capex regardless of which chip wins, albeit with a China/policy discount.

We remain constructive on the sector. Among the constituents, we like Memory, WFE and Foundry in particular. 

Nevertheless, for the whole index (SMH), our disagreement is with the price, not the business, which is why the thesis above is genuinely bullish on fundamentals even as the rating stays disciplined.

Top 10 constituent update                        

Holding	Weightage	Latest snapshot — technology & financials
NVIDIA (NVDA)	18.20%	Full-stack AI compute standard, Blackwell shipping, Vera Rubin next, with the CUDA and networking moat. FY26 revenue ~USD 216 bil (+65%), yet the stock is +4% YTD and in a correction at ~USD 193 among the the lower-valued AI mega-cap names on forward earnings. Read more: Nvidia Q127: Everything everywhere all at once
Taiwan Semiconductor (TSM)	9.00%	Indispensable leading-edge foundry: N2 (2nm GAA) reaches high-volume in Q4 2026 and CoWoS packaging is sold out through 2027 (>90% share). Q1 revenue USD 35.9 BIL with +58% net income, a 66% gross margin and HPC at 61% of sales. Read more: TSMC | The Kingmaker of the AI Era: Investment Opportunities Under the New Compute Hierarchy
Micron Technology (MU)	6.00%	Memory bellwether and thesis engine. HBM3E/HBM4 leadership, HBM4 ramping ~2x faster than HBM3E. FQ3 blowout (revenue USD 41.5 bil, EPS $25.11, ~83% cloud margin), HBM booked into 2028 and ~40% of revenue floor-priced. +325% YTD at ~$1,132 (~20x forward). Read more: Micron Q326: The Night Is Still Young
Broadcom (AVGO)	5.50%	Custom-silicon (XPU) and AI-networking champion, designing accelerators for Google, Meta, OpenAI and Anthropic. Record fiscal Q2: revenue USD 22.2 bil (+48%), AI semis USD 10.8 bil (+143%) with >USD 100 bil of FY27 AI line-of-sight. Read more: Broadcom Q2 26: Market frustration amidst absence of upward revision
Advanced Micro Devices (AMD)	5.40%	Credible second source: MI400 launches H2 2026 with 432GB HBM4 (~1.6x GB200), alongside EPYC and ROCm. Q1 data-centre revenue USD 5.78 bil (+57%); customers now include 6-gigawatt commitments from OpenAI and Meta; China MI308 rules are the live risk. Read more: AMD 1Q26: How far can it go
Applied Materials (AMAT)	5.40%	Broadest WFE toolkit including deposition, etch, CMP, metrology plus advanced-packaging tools, freshly elevated by the reconstitution. Re-entered the top 10 and jumped ~13% on the Micron read-through; US curbs on China-bound tools are the swing risk.
Intel (INTC)	5.10%	Turnaround and the only at-scale US leading-edge foundry hope: 18A (Panther Lake) across 200+ designs, backed by a USD 5 bil Nvidia stake and government support. Q1 delivered a surprise non-GAAP profit (~$0.20), though the foundry business continues to post losses. Read more: Intel Q1 26: Growth Trap
Lam Research (LRCX)	5.00%	Etch-and-deposition specialist with the most direct memory leverage (3D-NAND, HBM/DRAM deposition) and a growing advanced-packaging line. Record Q3 (revenue USD 5.84 bil, +24%; EPS +41%), the 2026 WFE view raised to USD 140 bil and ≥85% of FCF returned; same China overhang as peers.
KLA Corp (KLAC)	4.90%	Process-control near-monopoly, inspection and metrology that matter more as nodes shrink and yields dominate, on a high-margin service base. FQ3 revenue ~USD 3.4 bil (+11%) with a high-teens FY26 growth guide; completed a 10-for-1 split in May; quality and multiple both high, China the chief risk.
ASML Holding (ASML)	4.90%	The lithography chokepoint, sole EUV supplier, now ramping High-NA, with memory at 51% of Q1 system sales. Q1 net sales €8.8bn (+13%), a 53% margin, FY26 guide €36–40bn and a ~€38.8bn backlog. China down to ~20% of sales, a manageable, lower-margin headwind against a protected franchise. Read more: Every dollar spent on AI chips has to flow through this company

Source: Bloomberg Finance LP.,iFAST Compilation, Data as of 30 June 2026.

Neutral, TP at USD 717 with 13% upside

The earnings story is the best it has been in this cycle. Micron’s FQ3 blowout confirmed a memory inflection that is now partly contracted rather than purely cyclical; an AI-capex supercycle above USD 725 bil underwrites every layer of the basket; and June’s reconstitution tilted the index toward high-margin equipment. After a ~9% pullback from the all-time high, SMH trades at only 21.2x our FY28E earnings versus the 24x we deem fair, hence the headline upside.

We carry FY28E index EPS of 1,220, and at a held 24x that yields the $717 fair value. As such, we maintain semiconductor as 2.5 stars ‘Neutral’.

Table 2: SMH valuation

	FY25	FY26E	FY27E	FY28
PE Ratio (X)	57.97	34.31	26.43	21.23
Expected Earnings Growth		69%	29.80%	24.50%
Earning Per Share (EPS)	446.8	755	980	1220
Target Price (USD) (Fair PE of 24x)				717
Potential Upside (%)				13%

Source: Bloomberg Finance LP, iFAST Compilation, Data as of 30 June 2026.

Why 2.5 stars ‘Neutral’ when the upside is ~13% 

We like the asset and the earnings. We hold it at benchmark weight. We do not overweight, purely on valuation.

The current valuation multiple is full and struck on near-peak earnings. Paying 24x on FY28E estimates represents a generous through-cycle multiple for a basket whose FY27–28 profits lean heavily on memory and semiconductor equipment, sectors that the market historically capitalizes at much lower multiples near their peaks.

Furthermore, the broader tape remains highly fluid and concentrated. Even a blowout earnings print from Micron recently ended with chip stocks trading lower, and Nvidia, which constitutes roughly 18% of the fund is currently experiencing a correction. Amid rising ETF outflows and mounting (again) AI-bubble warnings, the more prudent approach is to add to positions on structural weakness rather than chasing strength at these levels.

The implied margin of safety remains relatively limited. Because small negative changes in either valuation or earnings can easily erase the headline reward, the risk-adjusted return looks ordinary.

Within this case, switching to Asia semiconductor (Global X Asia Semiconductor ETF) offers a more favourable risk-to-reward profile at current valuations.

Read more: Global X China Semiconductor ETF: Capturing Domestic Substitution and Memory Re-Rating

Key Takeaway

While we continue to see merit in the semiconductor landscape, but being selective is still crucial within the current backdrop.

That said, we maintain our recommendations as illustrated in the previous write-up:

For investors who are already positioned, we would maintain current holdings while directing new capital more selectively, including allocations to lower-correlation regions like Asia, to improve portfolio diversification.

For first-time investors, a phased entry approach via a Regular Savings Plan (RSP) remains a more disciplined way to build exposure as the sector transitions into a more mature phase of the cycle.

Remain invested, but avoid adding aggressively at current level of valuations.

Declaration:

For specific disclosure, at the time of publication of this report, IFPL (via its connected and associated entities) and the analyst who produced this report hold a NIL position in the abovementioned securities.

This research report was prepared with the assistance of artificial intelligence (AI) tools. iFAST Financial Pte Ltd does not rely exclusively on AI for content generation; the content of this report – including all investment theses, ratings, price targets and conclusions – has been independently reviewed and verified by the research analyst(s) to ensure accuracy and professional integrity.