Microsoft CEO Says AI Agents Drive Record Growth

Key Takeaways

• GitHub AI agents now start over 40% of pull requests, with Codex reaching 4 million active users, though service disruptions led to Copilot access limits and usage-based charging.
• 73% of developers use AI coding tools daily, with Anthropic's Claude Code named by 66% and Copilot growing to 50 million users; Chinese models handle over 30% of tokens on OpenRouter.
• Three AI valuation systems can agree on estimates like $1.6 million for properties, but shared assumptions about market areas may compromise accuracy.
• AI threatens entry-level jobs in call centers and data entry while creating opportunities in construction and manufacturing as data centers expand.
• SK Telecom's AI for Everyone campaign aims to democratize AI in South Korea through data centers, sovereign AI models, and one AI agent per citizen.
• AI microdramas risk aesthetic fatigue and cultural homogenization, according to Beijing and Hong Kong experts.
• Restraint is considered the most underrated skill in AI development, with experts advocating for honest progress over grand visions.
• India ranks third in Stanford's Global AI Vibrancy rankings, with its software talent and domestic market seen as AI advantages.
• Penn State's HAART theory reveals AI can simulate emotional connection but cannot truly experience human relationships.
• Japanese PM calls for global AI security cooperation in Paris, focusing on opportunities and challenges of frontier models.

GitHub reports AI agents now start over 40% of pull requests

GitHub data shows AI agents started over 40% of pull requests by September 2026, up from 17% in March. Microsoft CEO Satya Nadella says agent-based coding is driving record growth on its developer platform. OpenAI's Codex reached 4 million active users in April. GitHub faced over a dozen service disruptions lasting more than one hour since March due to the strain. The company began limiting Copilot access and will start usage-based charging in June.

Stack Overflow survey shows 73% of developers use AI tools daily

Stack Overflow's 2026 Developer Survey finds 73% of AI coding tool users use them daily, and 71% feel positive about them. Coding agent Anthropic's Claude Code is named by 66% of developers, while GitHub Copilot grew from 26 million users to 50 million between October 2025 and July 2026. Anthropic and OpenAI lead as preferred model providers. Chinese models now handle over 30% of tokens processed by U.S. companies on OpenRouter since February 2026. The most common AI use is generating code in familiar areas at 67%, followed by debugging at 61%.

Three AI home valuation systems may share the same flawed assumptions

Three AI valuation systems can produce similar home value estimates, such as around $1.6 million for the same property. However, agreement between the systems does not guarantee accuracy because they may rely on the same underlying assumptions about market areas and comparable properties. In dense markets like Miami, geographically close properties with similar features can actually compete in very different markets. Fannie Mae notes that competitive market areas should be defined by buyer alternatives rather than geographical proximity. Clean property data does not automatically mean the market analysis is correct.

AI threatens entry-level jobs but also creates new opportunities

Economists believe AI poses a particular threat to entry-level and repetitive work such as call center, data entry and junior copywriting jobs. Job growth in accounting, law and finance has been slow, while tech and media jobs have declined since 2022. Microsoft also offered buyouts to tens of thousands of employees. However, AI could also create opportunities in construction and manufacturing as companies build data centers and produce chips. The University of South Alabama and the Alabama Community College System offer AI training programs to help workers prepare.

SK Telecom launches AI for Everyone campaign in South Korea

SK Telecom launched a brand campaign titled AI for Everyone on November 10, 2026, aiming to democratize AI technologies and drive consumer adoption in South Korea. The initiative focuses on three pillars: building hyperscale AI Data Centres with global tech partners, advancing sovereign AI through its A.X K2 foundation model, and introducing the Everyone's AI personal agent platform targeting one AI agent per citizen. The campaign spans print, digital and video platforms. SK Telecom seeks to transition from telecommunications infrastructure into a high-margin AI platform and infrastructure exporter.

AI microdramas create aesthetic fatigue warns Beijing professor

Beijing professor Zhang Lanshan warned at a Macau forum that AI-generated microdramas cause aesthetic fatigue. He said audiences love standardised content but it risks stifling human creativity. Experts from Hong Kong institutions also raised concerns about cultural homogenisation and disinformation from large language models.

Restraint is AI's most underrated skill says industry expert

An industry expert argues that restraint is the most underrated skill in AI development. He says leaders must meet customers where they are and avoid overpromising on what AI can do today. He believes honest progress comes from small useful steps rather than grand visions that ignore real gaps.

Anand Mahindra says India's software talent is AI advantage

Anand Mahindra believes India's software talent and large domestic market could make it a major winner in the AI race. He highlighted India ranking third in Stanford University's Global AI Vibrancy rankings. He compared AI's future to smartphones where apps unlocked the platform's full potential.

Penn State professor studies human-AI relationship gap

Heather Shoenberger at Penn State co-created a Generative AI for All course taught since fall 2024. She also developed the HAART theory about how humans and conversational AI have a structural mismatch in relationships. Her research shows AI can simulate emotional connection but cannot truly experience relationships.

Japan PM calls for global AI security cooperation in Paris

Japanese Prime Minister Sanae Takaichi called for international coordination on AI security during a French-organized meeting in Paris. Foreign ministers from Group of Seven countries attended the gathering. The discussions focused on how advances in frontier AI models create both new opportunities and challenges.

Sources

• The biggest findings on AI agents, top languages and fastest-growing developer communities
• 6 AI and tech findings from Stack Overflow's 2026 Developer Survey
• 3 AI agents value a home similarly, could they be wrong?
• AI and your job: Threat or opportunity?
• SK Telecom brings artificial intelligence to everyday life through 'AI for Everyone' campaign
• AI microdramas cause ‘aesthetic fatigue’, professor tells Macau forum
• Restraint is AI's most underrated skill
• ‘It's the ecosystem, stupid!’: Anand Mahindra believes India's software talent could be its biggest advantage in AI race
• Faculty member addresses artificial intelligence practically and theoretically
• Japan PM calls for int'l coordination on AI security in French confab