Invest in ammunition manufacturers: Discovering potential in the artificial intelligence conflict
In the rapidly evolving world of artificial intelligence (AI), some of the biggest names in technology are grappling with the same disruptions as everyone else. Meanwhile, certain companies that play crucial roles in the AI hardware ecosystem are flying under the radar, offering potential investment opportunities.
Marvell Technology: A Fabless Chip Designer on the Rise
Marvell Technology, a fabless chip designer specialising in networking chips, has a strong foothold in data centres and cloud infrastructure. With shares around 24% below Morningstar’s fair value estimate, Marvell is considered significantly undervalued. Its broad portfolio in switching, processing, and optical chips positions it well to benefit from the demand for high-performance networking hardware generated by generative AI.
The market may be overlooking Marvell’s pivot into custom AI accelerators and cloud-focused chips, contributing to its undervaluation.
Taiwan Semiconductor Manufacturing Company (TSMC): The World's Leading Chip Foundry
TSMC, the world’s leading chip foundry, supports major AI hardware companies like Nvidia and AMD but trades at attractive valuations despite more than tripling in value over the past few years. Its undervaluation stems partly from market focus on fabless chip designers and premium GPU makers rather than on foundry service providers, even though TSMC is critical for manufacturing advanced AI chips.
Analysts highlight that some valuation metrics suggest TSMC remains "dirt cheap" relative to its foundational role in AI silicon production.
Micron Technology: A Crucial but Overlooked Player
Micron Technology, another undervalued player, mainly due to its focus on memory and storage rather than GPUs or ASICs, is another undervalued player. Its crucial role in supplying the memory backbone for AI applications is underappreciated, leading to a discount relative to other semiconductor stocks. As AI workloads increase demand for fast, large-scale memory systems, Micron's strategic significance is growing but not fully reflected in its market price.
The Race Among Tech Giants
Meanwhile, the so-called Magnificent Seven (Google, Microsoft, Amazon, Apple, Facebook, Alibaba, and Tencent) are engaged in an expensive battle among themselves to ensure they end up as one of the big winners rather than a loser from AI. Google Search, for instance, could become redundant if AI can consistently and reliably start to answer people's questions. Apple's position as the world's premier consumer-products company is under threat because it seems unable to weave AI into its products.
The Competitive Landscape
KLA, a dominant player in the wafer-inspection market, controls more than 50% of the market, and nearly half of the market for metrology equipment. Canon and Nikon, two serious competitors, can rival ASML's capabilities in lithography. Canon recently opened its first chipmaking equipment plant in 21 years, while Nikon started accepting orders for its DSP-100 Digital Lithography System last month.
TSMC makes all of the most advanced chips for Nvidia, Broadcom, AMD, and Qualcomm, as well as for the hyperscalers fighting the AI war - Microsoft, Google, and Amazon, and it has made chips for Apple for over a decade.
The Energy Consumption Conundrum
Data centres get through huge amounts of energy: Goldman Sachs estimates the world's data centres consume around 55GW of power, with this figure increasing 50% by 2027 and 165% by 2030. Companies like Calpine are signing agreements with data centre providers like CyrusOne to power new data centres, addressing this growing energy demand.
Additional companies like Nebius Group also show potential as undervalued AI infrastructure providers with strong growth prospects but carry higher risk due to cash burn and execution requirements.
In summary, these companies appear undervalued because the market tends to focus heavily on GPU manufacturers with obvious AI branding (e.g., Nvidia) and often underestimates the critical supporting roles of chip foundries, networking chipmakers, and memory suppliers essential for AI hardware ecosystems. Their stable growth potential, expanding AI-driven demand, and broad product portfolios suggest they may offer compelling investment opportunities that are currently priced below intrinsic long-term value.
- Despite Marvell Technology's strong position in data centres and cloud infrastructure, and its potential benefits from the demand for high-performance networking hardware generated by generative AI, its share price remains around 24% below Morningstar’s fair value estimate, suggesting it may be undervalued.
- TSMC, the world’s leading chip foundry, supports major AI hardware companies like Nvidia and AMD but is still considered undervalued, despite more than tripling in value over the past few years, due to market focus on fabless chip designers and premium GPU makers rather than on foundry service providers.
- Micron Technology, another undervalued player, is not fully appreciated for its crucial role in supplying the memory backbone for AI applications, leading to a discount relative to other semiconductor stocks as AI workloads increase demand for fast, large-scale memory systems.
- The market may be overlooking the potential of technology companies like Nebius Group, which show potential as undervalued AI infrastructure providers with strong growth prospects but carry higher risk due to cash burn and execution requirements.
