TOPT: Plenty Of AI, Just The Wrong Layers

iShares Top 20 US Stocks ETF (TOPT) received a Hold rating: while its total-return outlook is positive, it is not expected to outperform S&P 500 alternatives. The fund is highly concentrated, with roughly 71.5% of assets in its top 10 holdings and about 58% AI exposure, while its market-cap methodology overweights mega-cap names and misses equipment, memory and optical-interconnect opportunities. Limited sector diversification, including no industrials, utilities or materials exposure, constrains potential alpha.

Analysis

The relevant question is not whether mega-cap AI compounds, but whether incremental AI capex broadens beyond compute platform owners. A cap-weighted vehicle captures the most liquid beneficiaries but systematically dilutes exposure to the higher-operating-leverage parts of the buildout: wafer-fab equipment, HBM memory, advanced packaging, and optical connectivity. If hyperscaler capex guidance remains intact through the next earnings cycle, these bottleneck categories should deliver faster estimate revisions and relative multiple expansion than the largest index constituents.

The near-term risk is that AI spending increasingly becomes a zero-sum share battle among cloud platforms rather than a broad infrastructure cycle. In that case, megacaps retain defensive appeal while MU, COHR, LITE, and semiconductor-equipment names face sharper drawdowns because their valuations embed sustained order visibility. The key falsifier over 1-3 months is a reduction in aggregate hyperscaler capex guidance or evidence that GPU utilization and enterprise inference demand are not converting into new networking, memory, and foundry orders.

Consensus may be underestimating the portfolio-construction issue: concentration is not inherently bearish when earnings leadership is narrow, and broadening has repeatedly been delayed in prior AI rallies. The tactical signal is therefore conditional rather than an outright bearish call on TOPT: relative underperformance becomes more likely only once supplier order books and earnings revisions accelerate faster than the mega-cap cohort. Over 6-18 months, power availability, data-center construction constraints, and grid equipment shortages could create a second AI investment leg in utilities and electrical equipment that cap-weighted technology exposure largely misses.

Key Decisions for Investors

• Use TOPT only as a beta allocation, not dedicated AI exposure; for the next 1-3 months, prefer a relative-value basket long SMH and short TOPT if semiconductor earnings revisions continue to exceed S&P 500 revisions. Size modestly because a broad risk-off event favors mega-cap liquidity and can widen the spread against the position.
• Build a staged long basket in AMAT, LRCX, KLAC, and MU ahead of the next hyperscaler and semiconductor earnings cycle; these are higher-beta expressions of sustained AI infrastructure spend. Add only after order/backlog commentary confirms demand beyond a single customer, and exit if capex guidance is cut or memory pricing momentum reverses.
• Monitor COHR and LITE as high-risk, high-reward optical-interconnect alerts rather than core positions. Initiate only on independently confirmed 800G/1.6T order acceleration; the upside is substantial if networking becomes the next bottleneck, but customer concentration and execution risk warrant smaller sizing than equipment exposure.
• For a 6-18 month diversification sleeve, screen ETN, PWR, VRT, and CEG for AI-related power and data-center backlog acceleration. This is the cleaner hedge against a rotation from chip spending toward physical infrastructure; falsify the thesis if data-center project timelines slip or utility interconnection queues fail to convert into contracted load.