The AI model landscape was reshaped this week by a series of major releases that ignited a fierce price war. OpenAI launched its GPT-5.6 family, xAI debuted Grok 4.5, and Meta Platforms introduced Muse Spark 1.1, all with aggressive pricing that dramatically lowers inference costs. New flagship models are now priced between $4 and $6 per million output tokens, a steep drop from the $25 to $50 range of previous top-tier models from competitors like Anthropic. This strategic shift is set to accelerate AI adoption and intensify competition for high-volume enterprise customers.
Underpinning the software advancements is an unprecedented wave of infrastructure investment. A new forecast indicates that Amazon, Microsoft, Alphabet, and Meta Platforms will collectively spend approximately $725 billion on capital expenditures in 2026, a 77% increase from 2025, with a massive focus on data centers and custom AI chips. On the supply side, TSMC reported a 77% surge in quarterly profit fueled by AI processor demand and committed an additional $100 billion to its US operations in Arizona, bringing its total planned investment there to $265 billion.
Investor confidence in the AI sector reached new heights, from established platforms to emerging startups. Databricks announced it has signed a term sheet for a strategic funding round at a staggering $188 billion valuation, led by investor Coatue. In parallel, the AI agent space saw a surge in capital, raising $1.8 billion in July alone. Investment firms like Sequoia Capital were notably active, leading multiple deals, while companies such as Harvey AI secured significant rounds, underscoring a market trend towards specialized agents that deliver clear business value.
AI's integration into everyday products and enterprise operations also accelerated. Google began rolling out AI Mode integrations with third-party apps, allowing users to perform tasks with services like Instacart and Canva directly within search results. This move transforms Google Search into a proactive task-completion platform. In the cybersecurity realm, Microsoft revealed that its internal AI systems were instrumental in detecting a record 570 vulnerabilities addressed in its July security patch, highlighting AI's growing role in strengthening digital defenses.
The global AI landscape was also in focus as the World AI Conference (WAIC) opened in Shanghai, showcasing over 300 new products and a strong emphasis on practical applications. Chinese AI firm MiniMax unveiled its new flagship multimodal model, M3, capable of handling a 1 million-token context window. The event also saw 29 nations sign an agreement to establish the Shanghai-headquartered World Artificial Intelligence Cooperation Organization (WAICO), signaling a coordinated effort to shape international AI governance and cooperation.