Leading players in the artificial intelligence industry, including OpenAI, Anthropic, Google, and Microsoft, have significantly increased their lobbying budgets in the US during the first half of 2026. Simultaneously, Amazon is undergoing a massive reorganization of its AI assets, shifting from a wide range of models toward deep research into frontier systems.
What Happened
In the first half of 2026, US lobbying expenditures reached new highs. OpenAI directed $2.22 million, nearly doubling its previous figures, while Anthropic tripled its spending to reach $3.53 million. Google and Microsoft also set new historical records for lobbying activity. In parallel, Amazon announced the phasing out of development for many models in the Nova lineup, including the flagship Premier and Omni versions, as well as the Reel and Canvas media generators, redirecting resources to a new Frontier Model Research (FMR) group led by Peter Abbill.
Context
Current corporate activity is aimed at influencing key aspects of the industry's future: data center construction rules, model audit standards, and the regulation of open weights models. A fundamental shift is occurring: the market is moving from a stage of technological racing to a stage of regulatory struggle and strategic positioning.
Why It Matters for the Industry
For the industry, this signifies a transition from experimenting with a wide array of specialized tools to consolidating resources around the creation of ultra-powerful Frontier Models. The intensification of lobbying indicates that control over audit standards and infrastructural rules is becoming a critical factor for competitiveness. Furthermore, Amazon's actions signal a potential reduction in support for broad cloud APIs in favor of deeper fundamental research.
Why It Matters for Users
Users may encounter changes in the availability and support of current Amazon Nova cloud models, which may require a reassessment of application architectures. At the same time, the emergence of agentic systems, such as Perplexity's Personal Computer for Windows 10 and 11, marks a transition from simple chatbots to full-fledged AI assistants capable of directly managing local files and Microsoft Office applications.
What Remains Unknown / Limitations
There is a conflict of interest within the industry regarding regulation: Anthropic advocates for strict control over open weights models, while Microsoft maintains a more liberal approach.
Sources
Author
Look at AI, Editorial Team
