The Stack — June 06, 2026

Today's technology briefing covers notable advancements and challenges across AI/ML, industry movements, infrastructure, and the developer world.

AI/ML

The Estonian Language Institute has introduced a "Propaganda Resistance" benchmark to evaluate large language models (LLMs) on their ability to resist Russian propaganda. This initiative is crucial in safeguarding public discourse from foreign influence. Anthropic’s Claude models, specifically the Opus 4.7 version, performed exceptionally well in this benchmark. Additionally, new CLI tool Lowfat has been introduced to reduce AI token costs by filtering unnecessary output, and Gemma 4 has released Quantization-Aware Training (QAT) models optimized for mobile and laptop efficiency.

Industry

Anthropic is moving towards a public listing, having filed confidentially for an IPO. The company recently raised $65 billion, indicating strong investor interest despite broader market skepticism about AI's returns. In another significant move, AirTrunk announced a $30 billion investment to build AI data centers in India, reflecting the country's growing appeal for AI infrastructure. Supabase has also raised $500 million in a Series F round, doubling its valuation to $10 billion, driven by the increasing use of AI tools.

Infrastructure

A massive data center project in Utah has been significantly reduced in size due to local protests over environmental concerns, particularly regarding water usage and the impact on the Great Salt Lake. Meanwhile, Google is reportedly investing heavily in SpaceX's xAI data centers, paying $920 million per month for compute capacity, highlighting a significant expansion in AI infrastructure capabilities. On the cybersecurity front, Cisco's SD-WAN faces a new 0-day vulnerability, challenging administrators to maintain secure network operations.

Developer World

Microsoft has open-sourced pg_durable, a PostgreSQL extension enabling durable execution of SQL functions, enhancing fault-tolerant workflows within databases. Additionally, a critique of the Conventional Commits standard has emerged, arguing that it misplaces focus by prioritizing commit type over scope, potentially leading to inefficiencies in software development processes. In programming languages, the paper "Transformers Are Inherently Succinct" has been recognized at a top AI conference, marking advancements in the understanding of transformer models.