AI Infrastructure
Portfolio

The portfolio’s core exposure hyperscaler capital expenditure held up through a genuinely hawkish macro repricing. Despite the Federal Reserve’s June Summary of Economic Projections and a shift toward a “higher for longer” rates regime, estimates suggest AI focused companies will invest more than US$500 billion in infrastructure during 2026, while separate forecasts place hyperscaler capex commitments near US$750 billion. Capex conviction, rather than rate sensitivity, remained the dominant driver during the quarter.

Global semiconductor industry sales are projected to reach approximately US$1.5 trillion in 2026, hitting that milestone earlier than previously expected. This confirms that the structural demand backdrop remains intact even as financing conditions have tightened.

The principal macro risk is therefore not end demand weakness, but valuation sensitivity to interest rates. A hawkish policy surprise or a credit event in private AI infrastructure financing could trigger a rapid sector re rating, not because the demand thesis is impaired, but because long duration multiples remain highly rate sensitive. With the June Fed projections delivering precisely that hawkish surprise, this is the key macro thread to monitor into the third quarter.

The quarter exhibited a genuine two act structure. The Philadelphia Semiconductor Index fell approximately 10% before rebounding sharply, and the index crossed the 10,000 level for the first time in June a full round trip from correction to fresh highs within the same quarter.

Dispersion beneath the index level strength was extreme. Semiconductor focused exposures materially outperformed the broader technology sector, with selected names posting exceptionally strong year to date gains, while NVIDIA fell approximately 23% from its May all time high amid institutional rotation and renewed concern about higher rates. The underperformance of the largest position in many AI infrastructure baskets relative to smaller cap peers is an important momentum rotation signal.

The bull case remains earnings backed rather than purely narrative driven. NVIDIA guided to quarterly revenue materially above consensus and subsequently reported approximately US$81.6 billion of revenue, representing roughly 85% year on year growth. Even so, some technical analysts view the current advance as a late stage or fifth wave move, characterised by lower volume and flattening momentum. That interpretation does not invalidate the fundamental thesis, but it does raise the importance of monitoring late cycle risk.

The broader structural point is that only around 25% of hyperscaler capex flows directly to chips; the remaining 75% is deployed into data centres, power, networking and cooling. Momentum concentrated in pure play GPU names has therefore begun to decouple from momentum across the wider infrastructure stack. Portfolio monitoring should continue to assess the cycle layer by layer compute, memory, networking, power, cooling and digital infrastructure rather than relying solely on semiconductor index strength.

AI Infrastructure Q2 2026: The Capex Revenue Gap Is Now the Central Risk Variable

The bull case for AI infrastructure has never depended on whether demand exists. The central question is whether monetisation can catch up with spending before investors lose patience. Q2 2026 was the quarter in which that gap became impossible to ignore.

The scale of spending is unprecedented. The four largest US hyperscalers are guiding to approximately US$725 billion of capex in 2026, up roughly 77% year on year from about US$410 billion in 2025. Amazon is the largest at around US$200 billion, Microsoft is near US$190 billion, Google is guiding to approximately US$175 185 billion, and Meta to US$115 135 billion.

Against that spending base, current AI related cloud revenue remains materially smaller. Google Cloud is running at roughly US$80 billion annualised, AWS near US$150 billion annualised, and Azure AI around US$37 billion annualised. On some estimates, the gap between hyperscaler AI infrastructure spending and ecosystem revenue is now approximately US$600 billion per year and it is widening in 2026 rather than narrowing.

The return on capital hurdle remains demanding. Assuming hyperscalers require a 25% return on AI specific capex, the industry would need to generate approximately US$169 billion of annual AI attributable revenue by the end of 2028. Current AI cloud revenue is estimated near US$150 billion annualised. That is a credible gap to close, but it is still a shortfall and one that equity markets are not yet fully pricing as risk.

This is increasingly a financing story, not only a spending story. Hyperscalers are leaning more heavily on debt markets to bridge the gap between AI capex and internal free cash flow, marking a structural departure from historically cash funded models. More than US$100 billion of hyperscaler debt had reportedly been issued by mid March 2026, compared with roughly US$80 billion during the whole of 2025.

Oracle is the clearest stress case. Its large compute agreement with OpenAI drove a substantial increase in capex guidance to approximately US$50 billion, creating a funding gap of more than US$27 billion. Oracle’s five year credit default swap spread has more than tripled since September, with trading volumes well above prior norms. Credit markets, rather than equity markets, appear to be where the first meaningful scepticism is emerging.

Why can the hyperscalers not stop, even if ROI remains uncertain? Pulling back carries its own strategic risk. The companies that build the largest and most efficient data centres first gain asymmetric advantages in GPU access, training speed and partnership economics. Hyperscalers are not spending US$725 billion because returns are already proven; they are spending because being short of compute is the one mistake none of them can afford to make.

That creates a coordination game dynamic. Rational individual behaviour continue spending can still produce a poor collective outcome if enterprise monetisation disappoints. Because four companies are making the same bet simultaneously, the downside is highly correlated. If enterprise AI adoption stalls, the capex stack is likely to re rate across the board rather than gradually.

The risk is therefore asymmetric. As the capex number rises, the bridge between spending and eventual ROI becomes more fragile. In more aggressive scenarios, industry capex could approach US$1.4 trillion by 2027, which would deepen the funding cycle rather than resolve the monetisation question.

Portfolio positioning should distinguish between layers of the stack. The picks and shovels layer chips, power, cooling and data centre REITs has the clearest near term revenue visibility because it is supported by signed capex commitments. These businesses are paid regardless of whether enterprise AI applications monetise successfully.

The application and software layer is where the US$725 billion ultimately has to convert into durable revenue, and it is therefore the layer most exposed if the gap does not close by 2027 2028. This is a useful lens for ECATS Momentum weighting: infrastructure layer momentum is currently backed by contracted spending, while application layer momentum still depends on monetisation that has not yet been fully proven.

Q3 watch list: hyperscaler earnings calls for any change in ROI language or capex discipline; further widening in Oracle style CDS spreads as a leading indicator of credit market concern; and whether Azure, AWS and Google Cloud AI revenue growth can remain in the current 48 123% year on year range while the capex base continues to grow faster than revenue.

Reading the Market’s Second Layer

A quiet surface over a loud market.

On the surface, June looked uneventful. The S&P 500 drifted to record highs and volatility stayed low. If you only watched the index, there was nothing to see.

We don’t only watch the index. We track equity risk on three levels: what the options market charges for protection on the index, what it charges on the individual companies inside it, and how options dealers are positioned — because their hedging either cushions the market or accelerates it.

The first two levels have precise, published measures. The VIX prices thirty-day volatility on the S&P 500 as a whole. Its newer companion, the Cboe S&P 500 Constituent Volatility Index (VIXEQ), applies the same calculation to the individual stocks inside the index, weighted by their size — in effect, what the market charges to insure the average large American company rather than the basket. The two can diverge dramatically, and the gap between them is itself an index: the Cboe Dispersion Index (DSPX). When constituent volatility (VIXEQ) is high but index volatility (VIX) is low, the arithmetic permits only one explanation — stocks are expected to move a lot, just not together. The degree to which they move together is correlation, and it can be read directly from these three numbers.

That is exactly what June priced, to a record degree. VIXEQ finished the month near its highs — the average large-cap stock was priced to move at roughly three times the volatility of the index containing it, a relationship at the widest levels in the data’s history. Implied correlation among S&P 500 constituents ended the month near the lowest ever recorded. The calm index was an accounting artefact: AI infrastructure names repricing higher, rate-sensitive sectors falling, the two cancelling in aggregate. A quiet surface over a loud market.

The third layer: how dealers are positioned

Price alone doesn’t tell us how a shock will behave once it starts — whether it will be absorbed or amplified. For that we watch options dealers’ aggregate gamma exposure (GEX): the degree to which dealers must buy into a rally or sell into a decline to stay hedged, as a mechanical consequence of the options positions they’ve written to the market. When dealer gamma is positive, their hedging flow leans against the market and dampens moves — a “pinning” effect. When it turns negative, their hedging flow leans with the market and can accelerate a move in either direction. This is not a market view; it is a structural fact about who holds which options, and it changes day to day.

June’s most instructive episode illustrated exactly why we track this alongside price. A macro scare lifted index volatility (the VIX) more than 40% in four sessions. Alarming, on its face. But VIXEQ barely moved — the price of insuring individual companies was essentially unchanged, meaning the spike was entirely a repricing of correlation, not of risk. Dealer positioning corroborated the same read: the move had the signature of a fast, mechanical correlation event rather than a genuine reassessment of company-level risk. That combination — VIX up, VIXEQ flat, dealer flow consistent with a technical unwind rather than fresh selling — told us the episode would likely pass rather than persist. It did, within days. We held positions and hedges rather than paying the market’s worst prices to reduce either. Knowing which volatility to respect and which to fade is where this work pays for itself, and dealer positioning is often the tie-breaker between the two.

Implied volatility and where the yield comes from

The yield side of the programme lives inside the same data. Options premium is priced off implied volatility, and June’s dispersion meant implied volatility on individual holdings — particularly across AI-power, grid, and infrastructure-adjacent names — sat well above the volatility implied by the index itself. Writing calls against selected individual holdings in that environment captures materially more income per unit of upside surrendered than writing at the index level would; we rank holdings by where their implied volatility sits relative to their own recent history before sizing any call-writing, and we scale the programme back automatically when dealer positioning turns unfavourable, since rich premium in that setting is compensation for gap risk rather than free income.

While the overlay’s contribution in any single month will vary, the framework pays over time in three ways. When single-stock premiums are rich relative to the index — as they were most of June — our call-writing shifts toward individual holdings, collecting more income for each unit of upside we give away.

The same imbalance makes index protection unusually cheap, so we buy insurance before the storm, not after. And dealer positioning gives us an early read on whether market structure will absorb the next shock or amplify it — disciplining both how large our hedges run and when we pause premium harvesting altogether.

Wide dispersion under a calm index is, in the end, a stock-picker’s market — the environment our strategies are built for. It is also one of the rare periods when caution costs almost nothing. We are positioned for both.