OpenAI closes $110B raise from Amazon, NVIDIA, SoftBank in largest startup fundraise in history @ $840B post-money
Summary
OpenAI recently closed a significant funding round, securing $110 billion in new investment at a $730 billion pre-money valuation from Amazon ($50 billion), NVIDIA ($30 billion), and SoftBank ($30 billion). This investment is intended to scale infrastructure and bring AI to a broader audience. The company also disclosed substantial user growth: weekly Codex users have more than tripled to 1.6 million, paying business users relying on ChatGPT for work exceed 9 million, and ChatGPT boasts over 900 million weekly active users with more than 50 million consumer subscribers. Concurrently, Anthropic is embroiled in a dispute with the Department of War over refusing terms for mass domestic surveillance and autonomous weapons, leading to a potential "Supply-Chain Risk" designation. Other developments include Sakana AI's Doc-to-LoRA/Text-to-LoRA for instant LoRA adapter generation, Alibaba's Qwen3.5 model expansion, and advancements in inference systems like vLLM ROCm attention backends and DeepSeek's DualPath architecture.
Key takeaway
For CTOs and VPs of Engineering evaluating AI adoption and infrastructure, recognize that the rapid pace of investment and user growth in AI necessitates continuous assessment of vendor partnerships and ethical guidelines. Your decisions on model deployment should balance performance and cost-efficiency with the evolving geopolitical landscape and potential supply-chain risks, especially concerning government contracts. Prioritize flexible architectures that can adapt to both commercial demands and regulatory pressures.
Key insights
AI industry growth is marked by massive investments, rapid user adoption, and escalating ethical and geopolitical tensions.
Principles
- Ethical stances can lead to significant geopolitical friction.
- User growth and monetization are accelerating across AI platforms.
Method
Sakana AI's Doc-to-LoRA/Text-to-LoRA uses a hypernetwork to generate LoRA adapters in a single forward pass, converting task descriptions or documents into adapter weights with sub-second latency.
In practice
- Consider Qwen3.5 for fast scraping/summarization and GLM5 for complex code refactors.
- Utilize vLLM ROCm attention backends for up to 4.4x decode throughput on AMD GPUs.
Topics
- AI Funding
- AI Ethics & Governance
- LoRA Adapters
- LLM Benchmarking
- Hardware Optimization
Code references
Best for: Investor, CTO, VP of Engineering/Data, AI Engineer, Machine Learning Engineer, AI Researcher
Related on AIssential
Editorial summary, takeaway, and curation by AIssential. Original article published by AINews.