VC says top AI frontier models absorb 80%+ of spending as Alphabet’s Gemini claims No. 1 ranking

A new AI ranking and commentary from the venture capital world are converging on a single theme: the biggest, most capable “frontier” models may be increasingly tied to the largest pools of money. In a report carried by Yahoo Finance, a16z partner Martin Casado argued that the best frontier models take “80% plus” of AI spending when measured by dollars.

The same piece says Artificial Analysis’s Intelligence Index has placed Alphabet’s Gemini 4 Carbon at the top of its standings. The report frames the ranking as a announcement that Gemini’s latest model iteration is performing at the highest level among the models tracked by the index.

Casado’s point, as described in the coverage, is less about overall spending totals and more about how that money is allocated across the model ecosystem. By weighting results by dollars, the argument suggests that the resources required to train, iterate, and scale leading systems tend to concentrate heavily among the most advanced models rather than spreading evenly across many smaller efforts.

Alphabet’s Gemini program is positioned in that narrative as a beneficiary of the scale implied by the funding argument. While the report does not provide additional technical details or performance breakdowns in the material reviewed, it links Gemini 4 Carbon’s No. 1 Intelligence Index placement to the broader debate about what spending levels translate into top-tier results.

The story also implicitly raises a competitive question for the AI sector: if the leading frontier models capture most of the spending, then the market may tilt toward a small number of well-capitalized labs. That could increase the gap between top performers and less-funded challengers, especially as model quality becomes increasingly driven by costly data, compute, and infrastructure.