AI Industry Knowledge Graph

Entities and relationships, as a table

The same data the graph above draws. Editorial snapshot, not live data; see how this data is compiled.

Entities

• OpenAI (company): Creator of GPT series and ChatGPT. Backed by Microsoft. Leading AGI lab.
• Anthropic (company): Safety-focused AI lab. Creator of Claude. Founded by ex-OpenAI team.
• Google (company): Parent of DeepMind. Created Gemini, TPUs, and the Transformer paper.
• Google DeepMind (company): Google's AI research lab. AlphaFold, AlphaGo, Gemini. Nobel Prize 2024.
• Meta AI (company): Open-source AI push. LLaMA, SeamlessM4T. Yann LeCun leads research.
• Microsoft (company): $13B invested in OpenAI. Copilot integrated across all products. Azure AI.
• NVIDIA (company): GPU monopoly for AI training. CUDA, H100/H200/Blackwell chips.
• xAI (company): Elon Musk's AI company. Creator of Grok. Built Memphis supercluster.
• Mistral AI (company): French AI lab. Mixtral MoE architecture. Efficient open models.
• Hugging Face (company): The GitHub of AI. Model hub, datasets, Spaces. 500K+ models.
• Perplexity (company): AI-native search engine. Real-time web answers with citations.
• Cohere (company): Enterprise AI for search, generation, classification.
• GPT-4o (model): OpenAI's flagship multimodal model. Text, vision, audio in one.
• GPT-5 (model): OpenAI's most capable model. Doctoral-level reasoning. Released 2025.
• o3 / o4 (model): OpenAI reasoning series. Chain-of-thought at inference time. ARC-AGI record.
• Claude 3.5 Sonnet (model): Anthropic's best model. Top coding, analysis, reasoning benchmarks.
• Claude 4 (model): Anthropic's next generation model with enhanced reasoning and safety.
• Gemini 2.0 (model): Google's multimodal AI. Native audio/video/image understanding.
• LLaMA 4 (model): Meta's open source model. MoE architecture, free for commercial use.
• Grok 3 (model): xAI's flagship model. Trained on Colossus 100K GPU cluster.
• Mixtral 8x22B (model): Mistral's sparse MoE model. Matches GPT-4 class at lower cost.
• DeepSeek R1 (model): Chinese open reasoning model. Shocked industry with GPT-4 performance at fraction of cost.
• Gemma 3 (model): Google's open source small model. Runs on single GPU.
• Transformer (technology): The "Attention is All You Need" architecture (2017). Foundation of all modern LLMs.
• RLHF (technology): Reinforcement Learning from Human Feedback. Makes models safer and more helpful.
• Mixture of Experts (technology): Sparse activation architecture. Run only subset of params per token. More efficient.
• RAG (technology): Retrieval-Augmented Generation. Grounds LLM answers in real-time data.
• AI Agents (technology): LLMs that can plan, use tools, browse web, write code, execute tasks autonomously.
• Multimodal AI (technology): Models that understand text, images, audio, video simultaneously.
• Chain-of-Thought (technology): LLMs think step-by-step before answering. Dramatically improves complex reasoning.
• CUDA (technology): NVIDIA's parallel computing platform. Every major AI model trains on CUDA GPUs.
• Constitutional AI (technology): Anthropic's safety method. AI trained with written principles instead of human labels.
• Sam Altman (person): CEO of OpenAI. Led ChatGPT launch. $7T chip initiative. Survived board coup.
• Dario Amodei (person): CEO of Anthropic. Former VP Research at OpenAI. Safety-first philosophy.
• Yann LeCun (person): Chief AI Scientist at Meta. Turing Award winner. Skeptic of current LLM path to AGI.
• Demis Hassabis (person): CEO of Google DeepMind. AlphaFold creator. Nobel Prize Chemistry 2024.
• Elon Musk (person): CEO of xAI and Tesla. Co-founder of OpenAI (left). Building Grok.
• Jensen Huang (person): CEO of NVIDIA. Turned CUDA into the backbone of AI. "AI factory" vision.
• Ilya Sutskever (person): Co-founder of OpenAI. Chief Scientist. Left to start SSI (Safe Superintelligence).
• AGI (concept): Artificial General Intelligence, AI that matches or exceeds human capability across all domains.
• AI Safety (concept): Research ensuring AI systems remain aligned with human values as they become more capable.
• Scaling Laws (concept): More compute + more data + bigger models → predictable performance improvements.
• Open Source AI (concept): Publicly releasing model weights. Meta, Mistral lead. Debate: democratic vs dangerous.
• Inference Compute (concept): Spending more compute at test time to improve answers. o1/o3 paradigm.
• Microsoft → OpenAI (funding): Microsoft invested $13B+ in OpenAI across multiple rounds. Azure is exclusive cloud partner.
• Amazon → Anthropic (funding): Amazon committed $4B to Anthropic. AWS exclusive cloud partner.
• SoftBank → OpenAI (funding): SoftBank led $40B round at $300B valuation in 2025.
• NVIDIA Investments (funding): NVIDIA invested in Perplexity, Cohere, Mistral and dozens of AI startups.