AI Compute Shortage May Push Workloads to Chinese Models

Compute constraints threaten AI economics

The artificial intelligence industry faces a looming infrastructure bottleneck that could reshape the competitive landscape, according to new analysis from investment firm CLSA. Bhavtosh Vajpayee, an analyst at the firm, warns that surging demand for compute resources is outpacing available supply, creating cost pressures that may fundamentally alter how companies deploy AI workloads.

The mismatch between compute demand and infrastructure capacity is already manifesting in token rationing—a practice where AI service providers limit the number of requests users can make. This constraint directly impacts the economics of AI startups and enterprises that have built business models around readily available, affordable inference capacity.

Cost pressures may redirect market share

As compute costs rise and availability tightens, companies may increasingly shift workloads to less expensive AI models to maintain operational viability. This dynamic could create an opening for Chinese AI providers, which typically offer lower-cost alternatives to Western models from companies like OpenAI, Anthropic, and Google.

The potential market share shift represents a significant strategic concern for U.S. technology companies that have invested heavily in AI infrastructure. Chinese models, while sometimes perceived as less capable on certain benchmarks, may prove attractive enough for cost-sensitive applications—particularly as enterprises seek to control AI spending amid tighter compute availability.

Why it matters

This analysis highlights a critical tension in the AI industry: the gap between ambitious deployment plans and physical infrastructure reality. For business leaders, it signals that AI strategy must account for supply constraints and cost volatility, not just model capabilities. The compute shortage could accelerate demand for more efficient models and alternative architectures, while potentially fragmenting the global AI market along cost and availability lines rather than pure performance metrics.

Infrastructure investment under scrutiny

The warning comes as major technology companies continue massive capital expenditures on AI infrastructure, including data centers and specialized chips. Microsoft, Google, and other cloud providers have committed tens of billions of dollars to expanding compute capacity, but Vajpayee's analysis suggests demand growth may still outstrip these investments in the near term.

For enterprises planning AI deployments, the compute wall implies a need for more sophisticated workload management—potentially running different tasks on different models based on cost-performance tradeoffs rather than defaulting to a single provider.

These details were first reported by CNBC during a Squawk Box Asia segment featuring the CLSA analyst.