Weekend Edition: Navigating Token Costs Through Federated Routing, SLMs, and Open-Weight Models Can Set the Course for Earnings Acceleration

Research evidence excerpt

Weekend Edition: Navigating Token Costs Through Federated Routing, SLMs, and Open-Weight Models Can Set the Course for Earnings Acceleration

J P M O R G A N North America Equity Research

26 July 2026

Software

Weekend Edition: Navigating Token Costs Through

Federated Routing, SLMs, and Open-Weight Models

Can Set the Course for Earnings Acceleration

As AI continues to be integrated across all layers of software and customers move Software - Large Cap / Mid & Small

their AI projects from pilot to production, the economics of the underlying LLMs Cap

and tokens have become a central focus for the buyers and sellers, as well as Samik Chatterjee, CFA AC

investors. We analyze what we consider to be some of the key topics related to (1-212) 622-0798

tokens and how they permeate across our coverage and software more broadly. samik.x.chatterjee@jpmorgan.com

Specifically, we see companies using a combination of cost controls to try to Arti Vula, CFA

manage the higher-cost-to-serve AI solutions, with direct strategies including (1-415) 315-5919

intelligent model routing, use of domain-specific LLMs and small language arti.vula@jpmchase.com

models, and use of open-weight models, among various other strategies. In this Jaiden R Patel

(1-212) 270-6953

weekend edition, we walk through the different levers we are seeing companies jaiden.patel@jpmchase.com

use to lower token costs, including the benefits provided by the strong

Brian Hyska

performance exhibited by open-weight models, which has been in focus recently. (1-212) 622-2883

brian.hyska@jpmchase.com

1. Federated model orchestration helping play defense against rising Mashu Nishi

inferencing costs (1-212) 622-0078

A growing number of software companies are implementing a federated, multi- mashu.nishi@jpmorgan.com