Google DeepMind Appoints New Head; Anthropic Reports First Operating Profit
Google DeepMind's leadership transition and Anthropic's financial milestone signal evolving dynamics and a push for profitability among top AI labs, alongside continuous model releases.
The story
The AI software sector experienced significant internal shifts this week, with Google DeepMind appointing Koray Kavukcuoglu as its new Senior Vice President and head of the AI unit on August 12. Kavukcuoglu succeeds co-founder Demis Hassabis, who now chairs the organization, a move signaling an intensified focus on accelerating model development and closing competitive gaps.
Meanwhile, Anthropic announced its first-ever operating profit of $559 million on $10.9 billion in Q2 2026 revenue, significantly ahead of internal projections, demonstrating a clear path to financial viability for a frontier AI lab. Concurrently, new models and tools continue to emerge, with OpenAI releasing a cybersecurity-focused model and a Linux desktop app, while DeepSeek launched new agent frameworks. The market also saw substantial funding for new ventures, underscoring sustained investor confidence despite the evolving competitive dynamics.
Who moved
Google DeepMind
What Changed: Koray Kavukcuoglu was appointed Senior Vice President and head of the AI unit, taking over from co-founder Demis Hassabis.
Consequence: This leadership change signals a renewed push to accelerate model development and close competitive gaps with other major labs.
Anthropic
What Changed: The company reported its first operating profit of $559 million on $10.9 billion in Q2 2026 revenue.
Consequence: This achievement demonstrates a clearer path to financial viability for a major frontier AI lab, ahead of internal projections.
OpenAI
What Changed: The lab released GPT-5.6-Cyber, a specialized model designed to enhance defensive cybersecurity capabilities, and a ChatGPT desktop app for Linux.
Consequence: This expands OpenAI's offerings into critical security applications and improves accessibility for a wider user base.
River AI
What Changed: The startup, co-founded by former xAI co-founder Igor Babuschkin, raised $1.1 billion in a seed funding round just two months after its inception.
Consequence: This significant early-stage investment highlights investor confidence in new personal AI agent platforms and high-profile founding teams.
US State Department
What Changed: It launched the Pax Silica AI Assistance Project, committing up to $50 million to expedite the trade of AI-related materials with allies.
Consequence: This initiative aims to strengthen global AI supply chain security and foster trusted partnerships in AI development.
New models
GPT-5.6-Cyber
Lab: OpenAI
What: A specialized model designed to enhance defensive cybersecurity capabilities, capable of finding zero-day vulnerabilities.
Use: For cybersecurity professionals and organizations to improve threat detection and incident response.
Gemini 3.7 Flash
Lab: Google DeepMind
What: An updated model with highlighted improvements in coding accuracy and agentic task gains.
Use: For developers seeking more capable and efficient models for software development and automated workflows.
Harness v0.1
Lab: DeepSeek
What: An open-source agent framework released in developer preview.
Use: Enables builders to develop and deploy their own AI agents with a structured framework.
V4-Pro
Lab: DeepSeek
What: An updated model featuring agent upgrades, flexible reasoning controls, and native OpenAI Responses API support.
Use: For developers building sophisticated AI applications requiring advanced reasoning and integration capabilities.
GLM-5.3
Lab: Z.ai
What: An open model optimized for coding, agentic work, and cybersecurity benchmarks.
Use: Targets developers needing robust models for secure and automated tasks in coding and cyber defense.
Market signals
Anthropic reported $10.9 billion in Q2 2026 revenue and its first operating profit of $559 million.
Implication: This suggests that some frontier AI labs are beginning to monetize their models at scale and achieve financial self-sufficiency.
OpenAI's public S-1 filing is expected mid-to-late August, with pre-IPO disclosures indicating $2 billion/month revenue and a projected $14 billion loss for 2026.
Implication: This indicates high revenue growth but also substantial investment in scaling and research, highlighting the capital-intensive nature of frontier AI development.
River AI, a two-month-old startup, secured $1.1 billion in seed funding.
Implication: This demonstrates continued investor appetite for early-stage AI ventures, particularly those with high-profile founders in the personal agent space.
PitchBook's Q2 2026 data showed AI companies attracted 87.5% of all US venture dollars.
Implication: This highlights an extreme concentration of capital into the AI sector, potentially at the expense of other technology industries.
What we'll be watching
- Datavault AI's Q2 2026 financial results conference call on August 19, 2026.
- Actuate 2026, Foxglove's conference for robotics and physical AI developers, from August 17-19, 2026.
- Ray Summit 2026, Anyscale's annual conference for ML engineers and AI researchers, from August 24-26, 2026.
- TechBBQ 2026, a major Nordic tech conference in Copenhagen, from August 26-27, 2026.
- OpenAI's public S-1 filing, expected mid-to-late August, which will disclose audited financials and offering structure.
- OpenAI DevDay 2026, the annual developer conference in San Francisco, on September 29, 2026.
Reporting + analyst voices: grounded via Google Search at publish time.