1. Hyperscaler depreciation charges are expected to more than double by 2029.

Analysts expect the hyperscalers to incur $255 billion in depreciation charges in 2026 on capex of $784 billion.
The depreciation charge is projected to rise to $581 billion in 2029 on capex of $1.16 trillion.

“AI simply does not benefit from network effects, economies of scale, and proprietary technology to the same extent as social media, e-commerce, or software. Fundamentally, the AI industry is more like the EV industry — innovative, transformative, but also fairly commoditized, capital-intensive, and not very profitable.”

Source: BCA research

2. AI boom versus internet boom.

Falling costs should lead to more demand for AI models. (Jevon’s paradox)
The risk for AI companies is that demand does not rise fast enough to offset the drop in prices, causing revenue to stagnate.
This is what happened to telecoms.
The amount of data transmitted over the internet has grown at an annualized pace of 42% during the past 25 years, but spending on internet equipment has stagnated because today’s fibre can carry a lot more data than in the past.

Below: Internet traffic boomed but equipment spending stagnated

3. “The drivers that once underpinned luxury's premium ratings are fading”, says MS.

Luxury has already gone through a hefty de-rating, yet many long-only investors are still sitting on the sidelines. Why?
According to Morgan Stanley, the issue is no longer valuation alone. Instead, investors are questioning whether the industry's medium-term growth has permanently changed.
The bank argues that the drivers that once underpinned luxury's premium ratings are fading. At the same time, consumers are increasingly embracing second-hand luxury, while spending is shifting towards areas such as wellness and longevity.
"The historical pillars underpinning luxury’s premium are being challenged, or increasingly normalizing, such as strong pricing power, powerful structural growth tailwinds (China and overall democratization of luxury), expanding margins and low perceived earnings volatility," write MS analysts.
That uncertainty over the sector's medium-term growth rate is, in Morgan Stanley's view, keeping long-only funds away despite cheaper valuations.
The bank sees little scope for multiple expansion in the next 12 months and believes investors are increasingly treating the slowdown as a structural shift rather than a cyclical dip.

4. Global ESS battery shipments reached 112GWh in Aug-26, up 100% y/y.

Global ESS battery shipment growth further accelerated, supported by resilient China demand and robust exports to the EU and RoW.
China remained the main growth engine, with domestic shipments rising nearly 200% y/y, while Chinese suppliers’ shipments to the EU and RoW increased 120% and 136% y/y, respectively. Supply remained highly concentrated among Chinese producers, which accounted for 97% of global shipments in 8M26, while demand was more geographically balanced, with overseas markets representing 54% of shipments and China domestic demand the remaining 46%.

5. Nuclear meltdown.

Its future remains bright, but investors who chased the rally are suffering.
Even with some stocks down by well over half from their peak, valuations mostly remain too high.
The technology and future demand are real, but the AI nuclear boom is looking like Tchernobyl.

Not a subscriber yet?