What’s New in Artificial Intelligence – Oktober 2025

What’s New in Artificial Intelligence – October 2025

October 2025 brings a series of pivotal developments in AI, spanning infrastructure deals, strategic shifts in AI sovereignty, and enterprise integration of smarter models.

🏗️ Infrastructure & Strategic Partnerships

• OpenAI & AMD strike massive compute deal
  OpenAI and AMD announced a landmark partnership in which AMD will provide 6 gigawatts of AI GPU compute over multiple generations, starting with an initial deployment of 1 gigawatt in late 2026. This agreement positions AMD as a strategic infrastructure partner in the AI arms race. 

• EU escalates push for AI sovereignty
  The European Commission revealed a draft “Apply AI strategy” aimed at reducing dependence on U.S. and Chinese AI technology. The plan emphasizes boosting homegrown AI tools in public administration, healthcare, defense, and manufacturing. 

🏢 Enterprise & Governance

• IBM and Anthropic partner for enterprise AI
  IBM and Anthropic announced a collaboration to integrate Claude into IBM’s software and enterprise tools, embedding governance, security, and efficiency into the AI development life cycle. 

• eBay offers ChatGPT Enterprise to 10,000 sellers
  eBay granted ChatGPT Enterprise access to a large group of its sellers, aiming to streamline listing creation, customer responses, and sales analytics. This move signals deeper integration of AI tools in small and medium business operations. 

🚀 Emerging Trends & Use Cases

• AI + Mini‑Satellites expand Earth observation
  The commercial Earth observation market, now valued at around $9 billion, is increasingly powered by mini satellites coupled with AI analytics—enabling fast insights in climate science, agriculture, disaster response, and more. 

• Sora 2 avatars make waves (and worries)
  OpenAI quietly launched Sora 2, a consumer-facing avatar system: users upload a face scan, speak a few prompts, and a persistent, realistic avatar is generated that can produce videos with synchronized speech. The launch reignited concerns around deepfakes, identity, and consent. 

✅ What October Tells Us

• Control over compute infrastructure is becoming a strategic battleground—partnerships like AMD–OpenAI may shift balance away from incumbents.

• Sovereignty in AI technology is now a clear policy objective, especially in regions seeking to reduce foreign dependencies.

• Enterprise adoption of AI is moving from experimentation to integration—embedding models into workflows and software stacks.

• Advances in AI use cases (satellite imagery, avatar generation) continue to expand the boundaries of how we interact with and depend upon AI.