Bainbridge Partners portfolio manager Chawkat Nammour explains why surging AI demand, tightening memory supply and hyperscaler capital intensity are reshaping equity markets.
Euphoria surrounding AI stock has propelled the S&P 500 to record highs over the past several months. Despite geopolitical tensions and rising inflation concerns, the stock market recently posted eight weeks of consecutive gains. Leading the rally are hyperscalers, which continue to attract record inflows as their market cap reaches new highs. One figure trying to navigate the current landscape is Chawkat Nammour, a portfolio manager at Bainbridge Partners, a London-based multi-manager platform.
Nammour runs the firm’s Strand fund, which invests solely in AI, aerospace, and engineering software equities. Despite its tri-focus, AI dominates much of Nammour’s current thinking surrounding portfolio positioning; assessing the current moment, he admits the sector is pulling in two opposing directions. “The positive side is that AI in general is just exploding; the demand is off the scale. Anthropic’s ARR is growing rapidly, and the aggregate and total token usage is growing exponentially.Opposingly, we are concerned about the capital intensity of hyperscalers.” A Morgan Stanley investment note from February expected capital-intensity metrics for these hyperscalers to exceed levels last seen during the dot-com boom, believing they could drive about 40% of total Russell 1000 cash capex over 2026-28, a figure north of $2tn.
The concentration of capital in one area of the market is “concerning,” according to Nammour. In the past six months, the ubiquity of AI has increased drastically; for companies, it’s no longer a software problem but a hardware one. A more granular focus on the supply chain has seen investors vastly increase their exposure to hardware enablers. This coincides with Nvidia launching the Rubin Ultra Platform in 2027, a semiconductor that will significantly replace the current base architecture.
To mark this shift, Nvidia are having to source chip components with better memory capacity. However, despite suppliers Micron and SK Hynix reaching $1tn in market cap and driving substantial gains across Asian equity indices in recent months, demand continues to fall short of the industry’s growing supply. Creating a “bottleneck” situation that poses a threat to AI development.
Nammour explains, “Memory supply is only growing about 20% annually, and it takes two to three years to add meaningful new supply. Demand is growing much faster than that, which is driving prices sharply higher.” Nammour’s holdings in SK Hynix make up a significant portion of his portfolio, because the demand is always going to be there, thanks to cloud companies. Yet, he recently sold his position in Nvidia because “memory is the dominant theme and will increasingly eat into the industry profit pool. We need supply and demand to even out.”
Nammour’s withdrawal from Nvidia reflects his growing bearishness towards the current, exuberant market state. He trimmed his gross exposure closer to 100, from a high of 170, because he feels “there’s a real blow-up risk in the market”, citing the highly leveraged ETF’s in semiconductors – The Roundhill Memory ETF, with exposure to leading memory-chip stocks, has been one of the best-performing non-US equity investments this year – and the greater fluctuation in stock prices “stocks that used to move 5% on earnings now move 20%.”
Whilst AI equities can be prone to fluctuations, aerospace, where Nammour also invests, he sees as far more stable. The highly consolidated makeup of the sector means there are only a few major aircraft manufacturers, engine makers and suppliers. Investing is relatively low risk; Nammour highlights a key point that drives his thinking surrounding aerospace equities: “Roughly 80% of the world has never flown on a plane. As global living standards continue to rise and people enter the middle class, flying is one of the first things they aspire to do.” Additionally, the lack of supply options “means the sector can be very profitable, as cycles are much longer and can last up to 15 years.”
Nammour has always focused on AI, aerospace, and engineering software when investing. An engineer by trade, “The standout feature when investing is identifying companies that are building hard-to-replicate technology”, he cites SiTime as a prime example of this thesis. SiTime designs and manufactures silicon precision timing devices that control time, data transfer and radio frequency of electronic devices. Nammour notes that SiTime was the first to build an electronic timing system that “did not rely on quartz crystals – silicon control improves over time and therefore SiTime, we believe, will eventually dominate a very large market.”
Nammour is constantly refining his research and identification process when looking for companies like SiTime to invest. AI sits at the core of the fund’s operations – from scraper tools developed to compile overnight news into a morning bulletin, to Claude-powered automation for financial modelling, and proprietary tone-tracking systems designed to monitor market perception and sentiment around stocks. Hedgeweek® has spoken to several managers who feel these ‘edges’ that provide great value now, will eventually be arbitraged out. Nammour agrees, stressing that many AI tools in the investment process have become saturated, “most don’t improve workflow and decision-making; when I invest, I am looking for cyclical inflections. I want tools to help me identify and validate those turning points much faster.”
