Bridgewater CIO says AI compute rationing to create a 2-tier economy

A collision of geopolitical resource competition and ravenous demand for artificial intelligence is creating a severe shortage of computing power that will lead to a two-tiered global economy, according to Bridgewater Associates Co-CIO Greg Jensen. The dynamic is fueling the largest capital expenditure boom since World War II and initiating a brutal repricing of the entire software industry.

"The only thing in the world growing faster than compute supply is compute demand," Greg Jensen said in a recent interview. He argued that this imbalance is already forcing a quiet "rationing" of AI resources through higher prices and access restrictions, a trend he expects to accelerate dramatically.

Jensen’s firm forecasts that by the end of 2026, the gap between the supply of AI computing power and the demand for it will be so vast it will be equivalent to "wiping out all of the AI build-out that’s happened in the last year." The market is already pricing in this disruption, with valuations implying that only 82 percent of current software companies are expected to survive the transition over the next decade.

This creates a "winner-take-all" dynamic where companies that can secure access to leading-edge AI models and the underlying hardware gain an insurmountable structural advantage. For investors, Jensen noted, understanding this bifurcation between the compute-haves and have-nots is the key to navigating the next phase of technological change.

A Post-WWII Capital Expenditure Boom

The twin forces of "modern mercantilism"—where nations weaponize choke points in supply chains—and the AI arms race have kicked off a global scramble for resources. Jensen frames this as the largest capital spending cycle since the 1940s, with predictable macroeconomic consequences: a short-term surge in profits for suppliers, followed by sustained higher inflation and interest rates as the cost of capital rises to meet the demand for new infrastructure.

This is not a theoretical exercise. The effects are already rippling through adjacent industries. In gaming, the cost of memory chips, driven up by demand from AI data center operators like Alphabet and OpenAI, is forcing console makers like Sony to consider price hikes that could push their next-generation hardware toward the $1,000 mark, according to industry memos. This directly reflects the inflationary pressure Jensen describes, where the AI build-out crowds out other sectors.

Software's 'Barnes & Noble Moment'

Jensen compares the current state of the software industry to the "Barnes & Noble moment" of 1998, when the bookseller suddenly realized it had to compete with a nascent Amazon.com. Today, AI is fundamentally changing how software is built, forcing a painful re-evaluation of the entire sector. The market is no longer pricing software-as-a-service (SaaS) companies for perpetual, undisrupted growth.

Instead, a "disruption risk premium" is being priced in. The assumption that 100 percent of major software firms would exist in a decade has been slashed to just 82 percent, a figure Jensen calls the "essence of disruption" entering the market. This is visible in the mortgage industry, where dozens of vendors are embedding AI agents for document processing and workflow automation, as seen in offerings from MeridianLink and LenderLogix. Traditional software providers in such fields now face a stark choice: integrate generative AI capabilities or be rendered obsolete by more efficient, AI-native platforms. The companies that cannot afford the exorbitant cost of compute or secure partnerships with model providers will be, in Jensen's view, on the losing end of a "structural generational gap."

This article is for informational purposes only and does not constitute investment advice.