From processing power to computer memory
Artificial Intelligence took the world by storm in 2025 and every company even remotely associated with the sector saw their stock price increase dramatically. Vocal supporters of AI, such as Palantir (PLTR) and Alphabet (GOOG), rose 135% and 65% respectively last year. The companies building the chips required to train and run AI models also shot up. Nvidia (NVDA), Broadcom (AVGO), Taiwan Semiconductor Manufacturing Company (TSM), ASML Holding (ASML) and Advanced Micro Devices (AMD) were all fantastic plays last year, and these companies are likely to remain in the spotlight for much of 2026 as well. Chip manufacturers are a major piece of the AI puzzle, but so too are the companies responsible for providing memory and data storage. Vast amounts of computing power need to be matched with vast amounts short- and long-term storage capacity. While deeply intertwined, chip fabrication and memory/storage manufacturing are separate businesses. Up until now, attentions have been firmly focused on the former.
While Micron Technology (MU) absolutely smashed through the ceiling last year, gaining 240% in 2025, the company’s share price remains relatively down to earth compared to some of the price-to-earnings ratios seen in the wider technology sector. Although not as complex of an industry, memory can be just as much of a bottleneck to production as chip manufacturing. Just as with chipmakers, memory producers are a very small group of companies. Samsung, Hynix and Micron make up the vast majority of the market, with very little manufacturing capacity found outside of these big three.
The bottlenecks do not stop there either. Chip manufacturing and memory are arguably the larger components, but the AI sector depends on far more. Data servers, cloud computing, cybersecurity and high-speed, high-bandwidth infrastructure all play their part, but so too do the end user applications, including IoT devices, automation and software. All the companies downstream of developments in artificial intelligence are likely to garner their fair share of investor attention at some stage. Just as market sentiment trickled down from gold, to silver, to other metals last year, the AI sector may well experience a similar process in 2026.
Energy markets and Small Modular Reactors
Another crucial part of the Artificial Intelligence sector is the energy required to run it. The training of AI models is an extremely power-hungry activity that requires consistent and reliable sources of electricity. While oil and gas have traditionally provided much of the power used in heavy industry, modern tech companies are typically looking for something more on the green side. Wind and solar are all well and good, but do not meet the consistency requirement of data centres and server farms. Nuclear is the path forward, and the field is projected to see a sharp rebound in 2026. Meta Platforms (META) recently announced a 20-year agreement to buy nuclear power from Vistra (VST), while also committing to help the development of small modular reactors (SMRs) with Oklo (OKLO) and TerraPower. A collaboration between X-Energy, another SMR developer, and Amazon Web Services is being established. Digital infrastructure company Equinix (EQIX) is in partnership with Rolls-Royce SMR in an effort to pursue clean energy for its AI-driven data centres. The list goes on.
After many years of stagnation and plant closures across the world, the nuclear industry is undergoing somewhat of a renaissance. The sector has traditionally relied on huge, state-funded plants that require decades of construction and commissioning. Many such plants are currently undergoing refits and upgrades, but in the short term, something more flexible is needed. Emerging sectors of technology need clean, reliable power, and they need it now. Small modular reactors mark a completely different approach to the matter. Production is intended to be streamlined and commercially viable, pushing out mass-produced individual components that can be assembled on site, as and when power requirements arise. Need more power? Add a second reactor, or even a third and fourth. Nuclear startup companies are springing up across the world, with largely aligned goals and design philosophies.
The more complex an industry, the more complex its supply chains. Just as with the AI sector, the nuclear sector is heavily dependent on a huge number of sub-industries, all of which contribute to the larger picture. The most obvious to come to mind is probably the uranium mining industry, directly accessible via several ETFs. Enrichment and fuel assembly manufacturers come next. Reactor pressure vessels and steam generators both require advanced forging facilities – a relatively limited field. Turbines and electricity generating systems; electrical integration and substations; exotic materials for radiation-resistant components; civil engineering; software; waste management… The nuclear industry has a deep and diverse supply chain, each component of which is likely to benefit from renewed interest in the sector as a whole.
This is the year for cryptocurrencies (for real this time)
If there is one word that could describe cryptocurrencies in 2025, it would be the word “disappointing”. Bitcoin may well have hit an all-time high in October of last year, but it is fair to say that most people were expecting a little more than what they got. Bitcoin topped out at around $69,000 in 2021, a target that would not even be doubled in the subsequent bull run, which saw peaks of around $125,000. BTC would end the year around 6% in the red, meaning it got outperformed by basically everything. A savings account would have been a better bet. Storing money under a mattress would have yielded better results. The wider crypto market fared even worse than Bitcoin did. The fact that such a performance occurred during a year where every major stock market and precious metal hit a record high, only adds insult to injury.
Institutional interest was certainly present. ETFs saw huge inflows, strategic cryptocurrency reserves took off, and the regulation side of things was far more positive compared to previous years. And yet, the entire market felt boring, uneventful, subdued. Tamed, some might say. This is the problem with institutional money. The crypto sphere is starting to get the recognition it always craved. It should have been careful with what it wished for.
If the GENIUS Act was a small step in the right direction, then the Clarity Act would have been a giant leap in comparison. A number of events were supposed to occur last year, but because of the government shutdown in the United States, they did not. The Clarity Act is not dead by any means, but like any major piece of legislation, it is at the mercy of powerful interests and partisanship. The bill would set the stage for what is acceptable and what is not, and under which regulatory body different practices would fall. This is essentially the green light for companies who have already built vast networks of decentralised financial infrastructure but are unable to flip the switch out of fear of legal action. A frustrating situation. The nonsensical hype is over, soon to be replaced by novel financial instruments that should be beneficial to all. Once the relevant legislation comes into effect, both in the US and globally, the flood gates will open, and cryptocurrencies may finally enter a new era.
