A critical divergence exists in the AI market: hedge fund exposure to semiconductor stocks is at record highs, yet the primary buyers of these chips—the Mag7 hyperscalers—are showing market weakness. This creates a precarious situation where the supply chain's valuation is detached from its end-customer strength.
Related Insights
Firms like OpenAI and Meta claim a compute shortage while also exploring selling compute capacity. This isn't a contradiction but a strategic evolution. They are buying all available supply to secure their own needs and then arbitraging the excess, effectively becoming smaller-scale cloud providers for AI.
Current M&A activity related to AI isn't targeting AI model creators. Instead, capital is flowing into consolidating the 'picks and shovels' of the AI ecosystem. This includes derivative plays like data centers, semiconductors, software, and even power suppliers, which are seen as more tangible long-term assets.
Hyperscalers are selling their own securities (stocks, bonds) to fund a massive CapEx cycle in physical infrastructure. The most direct trade is to mirror their actions: sell their securities and buy what they are buying—the raw materials and commodities needed for data centers, where the real bottlenecks now lie.
A year ago, stable giants like Microsoft and Amazon absorbed the risk of the AI compute build-out. Now, they've stepped back, and smaller players like Oracle and CoreWeave, along with chipmakers financing their own sales, have taken on that risk. This shift to less stable, more circular financing models reveals the bubble's underlying fragility.
Current AI investment patterns mirror the "round-tripping" seen in the late '90s tech bubble. For example, NVIDIA invests billions in a startup like OpenAI, which then uses that capital to purchase NVIDIA chips. This creates an illusion of demand and inflated valuations, masking the lack of real, external customer revenue.
Companies like NVIDIA invest billions in AI startups (e.g., OpenAI) with the understanding the money will be spent on their chips. This "round tripping" creates massive, artificial market cap growth but is incredibly fragile and reminiscent of the dot-com bubble's accounting tricks.
Swisher draws a direct parallel between NVIDIA and Cisco. While NVIDIA is profitable selling AI chips, its customers are not. She predicts major tech players will develop their own chips, eroding NVIDIA's unsustainable valuation, just as the market for routers consolidated and crashed Cisco's stock.
The AI bubble may be less about the technology's potential and more about financial structuring. Companies like CoreWeave exist partly to absorb the low-margin, high-capex business of running GPUs. This protects the high-margin profiles of hyperscalers like Microsoft, preventing their stock from being dragged down by less attractive data center economics.
The AI infrastructure boom is a potential house of cards. A single dollar of end-user revenue paid to a company like OpenAI can become $8 of "seeming revenue" as it cascades through the value chain to Microsoft, CoreWeave, and NVIDIA, supporting an unsustainable $100 of equity market value.
When capital flows in a circle—a chipmaker invests in an AI firm which then buys the investor's chips—it artificially inflates revenues and valuations. This self-dealing behavior is a key warning sign that the AI funding frenzy is a speculative bubble, not purely market-driven.
