PIMCO remains constructive on AI but is reducing exposure to expensive hyperscalers and much of the Magnificent Seven.
PIMCO’s Emmanuel Sharef is looking beyond the Magnificent Seven and further down the AI supply chain, where chips, cooling systems, cables, power equipment and critical minerals turn digital ambition into physical infrastructure.
Takeaways
-
PIMCO remains constructive on AI but is reducing exposure to expensive hyperscalers and much of the Magnificent Seven.
-
The fund is overweight Asia, where semiconductor, memory, optical, cooling, power and construction suppliers provide direct exposure to the physical AI buildout.
-
Rare earths and industrial metals are becoming part of the AI trade because data centres ultimately run on hardware, electricity and physical materials, not algorithms alone.
Moving From Silicon Valley to Asia’s Factory Floor
The next phase of the artificial intelligence trade may look less like a victory lap through Silicon Valley and more like a tour of Asia’s factory floor.
That is the message coming from Emmanuel Sharef, portfolio manager of . The flagship 60/40 strategy has nearly $19 billion under management and has outperformed 97% of its peers over the past three years, according to data compiled by Bloomberg.
Beneath that traditional 60/40 exterior, Sharef is making a distinctly modern wager. The AI boom remains alive, but the next winners may not be the companies presently attracting the loudest applause.
PIMCO is not walking away from AI. It is moving away from the stadium’s most expensive seats.
Sharef says the fund is underweight most hyperscalers and much of the Magnificent Seven because valuations have become increasingly difficult to reconcile with the scale of the spending now required. The market has spent several years treating AI capital expenditure as an automatic conveyor belt to higher earnings. But the bill is becoming difficult to ignore.
Data centres do not rise from the ground on narrative alone. They need chips, memory, cooling, optical cables, power equipment, construction machinery and a small mountain of industrial metals. The hyperscalers must finance all of that before the promised productivity gains begin arriving through the front door.
That means heavier debt burdens, rising depreciation and more pressure on free cash flow. At some point, investors will want to see more than another upward revision to the capital expenditure budget. They will want evidence that those trillions are producing returns rather than merely producing larger data centres.
PIMCO’s answer is to avoid paying penthouse prices for the theme when much of the machinery can still be bought closer to the factory floor.
As Sharef told Bloomberg, investors do not need to own the market’s most expensive stocks to capture a major structural trend. That sounds obvious, but it runs counter to the muscle memory of an AI trade that has rewarded concentration and treated valuation discipline as unnecessary baggage.
The cleaner trade may sit further down the supply chain.
If the AI buildout keeps accelerating, suppliers across Asia collect the orders. If returns on that spending disappoint, the hyperscalers are left explaining the debt, depreciation and squeeze on free cash flow. In other words, the companies selling the picks and shovels may get paid before the prospectors discover whether there is enough gold in the ground.
That shifts the hunt toward semiconductor components, high bandwidth memory, cooling systems, cable interconnects, optical equipment, power supplies, construction machinery and industrial metals. These are not decorative extras hanging around the edges of the AI story. They are the physical ingredients without which the story never leaves the presentation deck.
Asia sits near the centre of that supply chain. The fund gained exposure last year through companies including , SK Hynix () and Taiwan Semiconductor Manufacturing (). Together, these businesses provide critical pieces of the AI production stack, from advanced foundry capacity to the memory required to keep accelerators moving data at industrial speed.
PIMCO remains overweight Asia because Sharef sees a combination of stronger earnings growth, more forgiving valuations and deeper exposure to the companies actually building the data centre ecosystem.
The market has spent much of the AI boom rewarding the architects. PIMCO is increasingly interested in the contractors.
There is also more to the Asia allocation than semiconductors. In China, the fund’s largest sector exposure is financials, where comparatively lower volatility provides ballast within the equity allocation. PIMCO has also been building positions in biotechnology and life sciences, where Sharef believes artificial intelligence could accelerate drug discovery and broaden the range of treatable diseases.
But the more intriguing extension of the thesis lies in materials.
, rare earths and other critical inputs are becoming the bridge between the digital AI narrative and the hard constraints of the physical world. Every new data centre requires vast amounts of electrical equipment, transmission capacity, cooling infrastructure and specialised components. Chips may provide the intelligence, but metals provide the skeleton and nervous system.
That gives Chinese resource extraction and materials companies a potentially important role. China remains deeply embedded in global rare earth supply chains, while export restrictions have reminded Washington that technological sovereignty is difficult to achieve when essential inputs remain concentrated elsewhere.
This is where the AI trade begins to bleed into industrial policy and geopolitics. The United States can subsidise fabrication plants, accelerate data centre approvals and throw capital at domestic supply chains, but it cannot manufacture geology. Alternative sources can be developed, although mines and processing facilities operate on a far slower clock than software.
The constraints are becoming easier to see. Power grids must be expanded. Cooling systems must become more efficient. Copper supplies must keep pace. Rare earth processing must be secured. The digital revolution is starting to collide with the stubborn realities of concrete, electricity and dirt.
For traders, that changes the shape of the opportunity. The original AI trade was narrow, obvious and increasingly expensive. The next phase is broader and more industrial, with returns potentially spreading through Asian semiconductors, memory, optical equipment, power infrastructure, construction machinery and materials.
That does not make the Magnificent Seven bad companies. It means great businesses can still become uncomfortable investments when valuations assume an almost flawless future. Meanwhile, the suppliers can enjoy accelerating orders without carrying the same burden of expectation.
The first leg of the AI boom rewarded the companies selling the dream. The next leg may reward the Asian companies manufacturing the reality.
That is the real PIMCO pivot. AI has not stopped being a technology trade, but it is rapidly becoming an industrial, power and materials trade as well. The market may still be staring at the marquee while the better risk-reward quietly migrates backstage.
Earlier last month from Bloomberg Aug 3
PIMCO 60/40 Fund Attracts $10 Billion Inflows on Asia Demand
Takeaways
-
Pacific Investment Management Co. mutual fund received more than $10 billion of inflows in the first half, showing investor demand for a traditional 60/40 split between stocks and bonds.
-
The Pimco Balanced Income and Growth Fund has more than doubled assets since the end of 2025 to $16.3 billion as of June 30, and returned more than 10% after fees in the first half.
-
The fund has benefited from a relatively higher “neutral equity posture,” and has beaten its own 60/40 benchmark, showing strength in stock and bond selection.
Pacific Investment Management Co. mutual fund that follows a traditional 60/40 split between stocks and bonds received more than $10 billion of inflows in the first half, showing investor demand for a strategy that has occasionally been called into question due to extreme moves in global markets.
Wealthy clients in Taiwan, Hong Kong, Singapore and mainland China are among top investors in the fund this year, after assessing risks tied to the Middle East conflict, according to Marcio Bogoricin, Pimco’s head of global wealth management for Asia excluding Japan.
The prolonged Iran war has caused volatility in oil prices, sparking inflation fears and increasing prospects for higher interest rates. That has reduced the appeal of bonds, undermining the notion that they would offer protection to investors in a 60/40 portfolio when stocks falter. Pimco’s fund, however, has been growing as it generates double-digit returns.
“We had a strong start to the year,” Bogoricin said in an interview. “In March, investors paused following the outbreak of the war,” but flows into the fund “continued to be strong.”
The Pimco Balanced Income and Growth Fund has more than doubled assets since the end of 2025 to $16.3 billion as of June 30. It returned more than 10% after fees in the first half, after posting a 21.65% gain last year.
The fund’s 60% stock allocation tapped artificial intelligence trades by investing in firms like Samsung Electronics Co., SK Hynix Inc. and Taiwan Semiconductor Manufacturing Co. last year, when their values were soaring, said Bogoricin. It began trimming those exposures earlier this year. More recently, it’s been looking for opportunities to invest in equipment suppliers in markets outside of the US, such as Japan, he said.
With more than $14 billion of net inflows since the start of 2025, the fund led the runner-up, , by almost three times, according to the latest figures from Morningstar Inc., which tracks 250 funds with similar asset breakdowns.
The fund has benefited from a relatively higher “neutral equity posture,” said Sam Hui, a senior analyst at Morningstar. “It has also beaten its own 60/40 benchmark, showing strength in stock and bond selection.”