Chamath Palihapitiya backs Meta’s aggressive AI pricing strategy after “Muse Spark 1.2” benchmarks, says Meta is going for “scorched earth”

Meta Platforms is facing a growing test of whether cutting the price of AI services can be paired with strong performance. In a recent public commentary, venture capitalist Chamath Palihapitiya said Meta is taking what he described as a “scorched earth” approach to AI pricing, arguing the company’s strategy has been validated by fresh benchmarks for its $0.69 model, Muse Spark 1.2.

Palihapitiya’s point, as reported by Yahoo Finance, hinges on performance metrics that he said place Muse Spark 1.2 among the top five on “Vals,” a benchmark referenced in the report. The framing is notable because it treats price as part of competitive positioning, not only as a marketing tactic.

In the same account, Palihapitiya characterized the broader move as a pricing strategy designed to reshape the market. While the commentary did not provide granular details on how Meta set its $0.69 price, the thrust of the argument is that cheaper inference, paired with benchmark-relative strength, can win customers faster than incremental improvements alone.

Meta’s AI business context matters here because the company has spent heavily on building out model capabilities and deploying them through consumer and developer-facing products. In general terms, AI providers live at the intersection of two constraints, cost to run models (inference and infrastructure) and demand for quality outputs (which can depend on model size, training, and tuning). When providers lower prices, they often need to ensure their systems can maintain output quality and compute efficiency at scale.

The report’s mention of Muse Spark 1.2 also highlights how quickly the AI industry is moving from “can it work” to “can it work economically.” For many users, the usable unit is no longer only a model name, but the price per request, responsiveness, and the practical quality of results for specific tasks. A model presented at a fixed dollar price point can become a default choice for developers testing prototypes and for teams evaluating vendors.

Meta has not, in the material referenced in this coverage, disclosed additional details that would let an outside observer independently verify the benchmark setup, the evaluation criteria, or what exact costs the $0.69 figure is intended to represent (for example, whether it corresponds to a per-token, per-request, or usage-tier construct). Without those particulars in the published commentary, the debate remains focused on what Palihapitiya says the benchmarks show rather than on independently audited methodology.

Still, Palihapitiya’s remarks add fuel to a wider industry argument: whether aggressive pricing can force competitors to accelerate cost reductions. If customers come to expect lower unit economics while performance stays strong, that can pressure other AI providers to match prices, negotiate distribution deals differently, or offer bundled tiers.

Investors and product teams will likely watch whether Meta further clarifies the commercial logic behind Muse Spark 1.2, including how pricing maps to usage and what performance the company claims across different workloads. The next announcement to monitor, beyond any new benchmarks, is whether Meta’s pricing stance translates into measurable adoption in real products, such as increased usage by developers or expansion of AI features across its platforms.