One of the most persistent myths in investing is that buying at record levels is reckless. Historical data going back to 1950 tells a different story. Purchasing the S&P 500 at the exact moment it reaches new peaks has consistently delivered stronger one, three, and five-year forward returns than at any other moment. New heights are not a warning sign. They are, statistically speaking, one of the best entry points available.

Tim Cook will step down as Apple's chief executive on 1 September, closing a chapter that transformed the company from a $350 billion enterprise into a $4 trillion colossus. He will remain involved as executive chairman, handing the reins to hardware chief Jeff Williams. The transition comes at a delicate moment: Apple is widely regarded on Wall Street as having fallen behind its mega-cap peers in artificial intelligence and closing that gap will become the defining test of the incoming leadership.

Since hostilities broke out on 27 February, visible global oil inventories have shed 255 million barrels, bringing the total to 7,864 million barrels. The pace of depletion has accelerated sharply: April recorded the heaviest monthly draws since 2017, at nearly 11 million barrels per day, pushing cumulative losses since the war began to 474 million barrels. Flows through the Strait of Hormuz remain at roughly a tenth of normal levels. Even under optimistic scenarios that assume a partial Hormuz recovery by late April, Goldman Sachs analysts warn that all-time low inventory levels may prove unavoidable.

OpenAI has crossed a valuation milestone that would have seemed fantastical just a year ago: pre-IPO instruments now imply a market value of $1 trillion, up 163% since October 2025. The company is not alone at that altitude, Anthropic is reportedly approaching a similar figure ahead of a potential public offering, while SpaceX is said to be targeting a valuation north of $1.7 trillion. The world has never seen so many private companies simultaneously occupying territory once reserved for only the most established public giants.

NVIDIA chief Jensen Huang recently offered a framework for mapping the entire artificial intelligence economy as a five-tiered structure. At the base sits energy: AI data centres demand vast quantities of electricity, making nuclear and renewable power infrastructure strategically critical. Above those come chips, where companies like NVIDIA, TSMC, and Broadcom convert raw power into computing capacity. The third layer is infrastructure, the massive GPU clusters and cloud facilities that coordinate compute at scale, with players such as Oracle, CoreWeave, and Nebius building the backbone. Next are models, the large AI systems that process data and generate intelligence, where Meta, Microsoft, Amazon, and Alphabet compete intensely. At the top sit’s applications, the layer where AI generates tangible economic value: autonomous vehicles, enterprise software, robotics, and AI agents. Huang's core message is that AI has evolved from software into foundational infrastructure, analogous to electricity or the internet.

America's data centre expansion is running into serious headwinds. Satellite analysis by SynMax, combined with IIR energy data, finds that close to 40% of facilities scheduled to open in 2026 face delays of more than three months. The pipeline for 2027 is even more concerning, as projects representing roughly 50 gigawatts of planned capacity, equivalent to the output of 50 nuclear reactors, have yet to break ground as of April 2026. The culprits are structural: chronic shortages of specialist labour, gas turbines, and electrical transformers, compounded by permitting delays that are pushing labour costs up by as much as 30% in remote sites. The question of whether the much-hyped buildout can stay on track is becoming increasingly hard to answer with confidence.

Chegg built a profitable business helping students with homework and textbook rentals. Today its share price has collapsed by nearly 99%, hovering below $1 with market capitalisation of around $100 million, a fraction of where it stood at its 2022 peak. The cause is straightforward: tools like ChatGPT did not simply compete with Chegg's offering; they rendered it redundant. When students can obtain instant, high-quality answers at little or no cost, a subscription-based homework platform struggles to justify its existence. Chegg's fate is being watched closely as a preview of what AI disruption can do to a business with speed and finality, and the broader question it raises is which category faces a similar reckoning next.