In our prior piece 'AI's Original Sin: Is NVIDIA the stimulant AI can’t quit?', Dolphin Research argued that the key shift for the AI supply chain in 2025 is the extreme upstream capture of value, which has thrown the system out of balance. This imbalance is reshaping incentives and risk-taking across the stack.

With FOMO running high, AI spend keeps climbing, while rivalry among core players has become more explicit. Capital allocations are increasingly concentrated at the top of the chain.

As the internet enters the AI era, regardless of how shifting inputs reshape bargaining power along the value chain, one ultimate question remains. After front-loaded investment in heavy capex, deep tech, and top talent with an approx. five-year depreciation cycle, is this ultimately a bubble?

Marked by the release of ChatGPT at the end of 2022, the AI frenzy over the past three years has swept through every sector, from computing power, storage, networking, manufacturing, power infrastructure, software applications, and even to edge devices. Each direction has been repeatedly hyped. However, by the end of the third anniversary, when the pillars of AI infrastructure announced unprecedented AI mega infrastructure projects around the third-quarter reports, the market suddenly seemed to lose its soul, starting to worry that AI investments might be a bubble. Some in the industry have made a fortune, while others have lost so much that they resort to frantic fundraising. Where exactly is the problem? The financing is dazzling and bizarre, why? This article...