The atlas, in full
Fable 5 · data curated through July 2026
Every company, event and person in the Five Layer AI Cake, as text. The interactive version is on the front page; this page exists so the dataset can be read, quoted and cited without running any code.
How a company is placed
A company sits on the layer where it captures value, not where it operates. NVIDIA runs enormous cloud and software businesses and still belongs to Silicon, because that is where the money is made. The rule is what makes this a map rather than a directory, and it is the one editorial decision everything else follows from. A company that captures value on two or more layers is a vertical integrator, and those are tracked separately because spanning layers is a different game from competing on one.
The five layers
| Layer | Short | What it is | Companies |
|---|---|---|---|
| 1 · Silicon | GPUs, TPUs, ASICs | Raw compute. Who makes the metal that every layer above depends on. | 18 |
| 2 · Cloud & Compute | Hyperscalers, neoclouds | Where the chips live. Data centers, GPU clusters, and the contracts that route demand. | 23 |
| 3 · Foundation Models | The frontier labs | The model itself — architecture, weights, capabilities. The narrow funnel the whole stack aims at. | 39 |
| 4 · Tooling & Orchestration | SDKs, agents, frameworks | The connective tissue — how developers wire models into real software. | 24 |
| 5 · Applications | End-user products | The surface users actually touch. Where attention, subscriptions, and enterprise contracts accumulate. | 46 |
All 103 companies, by layer
Influence is a subjective 1–10 score. Use it as an ordering, not a measurement.
Silicon — 18 companies
| Company | Founded | Layers | Influence | Why it is here |
|---|---|---|---|---|
| NVIDIA | 1993 | Silicon | 10/10 | GPU monopoly on AI training. CUDA moat. |
| Google Cloud | 2008 | Cloud & Compute · Silicon · Foundation Models · Applications | 9/10 | TPUs + Gemini + Workspace. Only full-stack vertical. |
| TSMC | 1987 | Silicon | 9/10 | Fabs every leading-edge AI chip on Earth. |
| AMD | 1969 | Silicon | 7/10 | MI300X, ROCm — the credible challenger. |
| Broadcom | 1991 | Silicon | 7/10 | Networking silicon for every hyperscaler. Tomahawk switches, custom XPUs. |
| Arm | 1990 | Silicon | 6/10 | Architecture under every phone + edge device. |
| Groq | 2016 | Silicon · Cloud & Compute | 6/10 | LPU — record-setting inference latency. |
| Huawei (HiSilicon) | 1991 | Silicon · Cloud & Compute | 6/10 | Ascend 910C/D — China's domestic GPU answer to NVIDIA under export controls. |
| Qualcomm | 1985 | Silicon · Tooling & Orchestration | 6/10 | Bought Modular to fight CUDA with software. Silicon without a dev stack does not sell. |
| Tesla (Optimus, FSD) | 2003 | Applications · Silicon | 6/10 | Dojo chips + Optimus + FSD — an AI company hiding in a carmaker. |
| Arista Networks | 2004 | Silicon | 5/10 | Data-center Ethernet. The anti-InfiniBand bet for AI clusters. |
| Cerebras | 2016 | Silicon · Cloud & Compute | 5/10 | Wafer-scale engine. Training alt to NVIDIA. |
| Intel | 1968 | Silicon · Cloud & Compute | 5/10 | Gaudi chips, foundry ambitions; fell behind in AI. |
| SMIC | 2000 | Silicon | 5/10 | China's foundry champion, shipping 7nm without EUV. |
| Corning | 1851 | Silicon | 4/10 | Fiber optic cable for sea-cable and data-center builds. 170-year-old picks-and-shovels. |
| Lightmatter | 2017 | Silicon | 4/10 | Optical interconnect — photonic compute fabrics. Cluster-scale bet. |
| Tenstorrent | 2016 | Silicon | 4/10 | Jim Keller’s RISC-V AI chips. Open-source ISA bet. |
| SambaNova | 2017 | Silicon · Cloud & Compute | 3/10 | Reconfigurable dataflow architecture. |
Cloud & Compute — 23 companies
| Company | Founded | Layers | Influence | Why it is here |
|---|---|---|---|---|
| Microsoft Azure | 2008 | Cloud & Compute · Foundation Models · Applications | 10/10 | OpenAI infra. Copilot everywhere. |
| AWS | 2006 | Cloud & Compute | 9/10 | Cloud hyperscaler. Bedrock model gateway, Trainium/Inferentia chips. Anthropic partner. |
| Google Cloud | 2008 | Cloud & Compute · Silicon · Foundation Models · Applications | 9/10 | TPUs + Gemini + Workspace. Only full-stack vertical. |
| CoreWeave | 2017 | Cloud & Compute | 7/10 | Neocloud darling. Pivoted from crypto to AI. |
| Databricks | 2013 | Cloud & Compute · Tooling & Orchestration · Foundation Models | 7/10 | Acquired MosaicML ($1.3B). Enterprise AI plumbing. |
| Equinix | 1998 | Cloud & Compute | 7/10 | 260+ IBX colos + 2,400+ interconnects. Where clouds peer with each other. |
| Vercel | 2015 | Tooling & Orchestration · Cloud & Compute | 7/10 | AI SDK, v0, AI Gateway. Frontend-first infra. |
| Constellation Energy | 2022 | Cloud & Compute | 6/10 | Restarting Three Mile Island for Microsoft. America's largest nuclear operator. |
| Digital Realty | 2004 | Cloud & Compute | 6/10 | 300+ data centers globally. The largest AI-era landlord. |
| Groq | 2016 | Silicon · Cloud & Compute | 6/10 | LPU — record-setting inference latency. |
| Huawei (HiSilicon) | 1991 | Silicon · Cloud & Compute | 6/10 | Ascend 910C/D — China's domestic GPU answer to NVIDIA under export controls. |
| Cerebras | 2016 | Silicon · Cloud & Compute | 5/10 | Wafer-scale engine. Training alt to NVIDIA. |
| Intel | 1968 | Silicon · Cloud & Compute | 5/10 | Gaudi chips, foundry ambitions; fell behind in AI. |
| Oracle Cloud | 2016 | Cloud & Compute | 5/10 | Stargate partner. Back-door into AI infra. |
| Snowflake | 2012 | Cloud & Compute · Tooling & Orchestration | 5/10 | Cortex LLM functions. Enterprise data gravity. |
| Together AI | 2022 | Tooling & Orchestration · Cloud & Compute | 5/10 | Open-model inference cloud. |
| Vantage Data Centers | 2010 | Cloud & Compute | 5/10 | Hyperscale builds. DigitalBridge-backed. 1GW+ campuses. |
| Fireworks AI | 2022 | Tooling & Orchestration · Cloud & Compute | 4/10 | Low-latency inference for open models. |
| Kairos Power | 2016 | Cloud & Compute | 4/10 | Small modular reactors (SMRs). Google signed for 500MW. Deployment 2030. |
| Lambda Labs | 2012 | Cloud & Compute | 4/10 | GPU cloud for researchers and startups. |
| Replicate | 2019 | Tooling & Orchestration · Cloud & Compute | 4/10 | Run any model via HTTP. |
| X-energy | 2009 | Cloud & Compute | 4/10 | High-temp gas-cooled SMRs. Amazon backing + $500M Series C-1 2024. |
| SambaNova | 2017 | Silicon · Cloud & Compute | 3/10 | Reconfigurable dataflow architecture. |
Foundation Models — 39 companies
| Company | Founded | Layers | Influence | Why it is here |
|---|---|---|---|---|
| Microsoft Azure | 2008 | Cloud & Compute · Foundation Models · Applications | 10/10 | OpenAI infra. Copilot everywhere. |
| OpenAI | 2015 | Foundation Models · Applications · Tooling & Orchestration | 10/10 | ChatGPT, GPT-4, Sora, o-series. Defined the category. |
| Anthropic | 2021 | Foundation Models · Applications · Tooling & Orchestration | 9/10 | Claude, MCP, constitutional AI. Safety-first frontier lab. |
| Google Cloud | 2008 | Cloud & Compute · Silicon · Foundation Models · Applications | 9/10 | TPUs + Gemini + Workspace. Only full-stack vertical. |
| Google DeepMind | 2010 | Foundation Models | 9/10 | AlphaGo → Gemini. Research arm of Google AI. |
| Hugging Face | 2016 | Tooling & Orchestration · Foundation Models | 8/10 | The GitHub of ML. 1M+ models hosted. |
| Meta AI | 2013 | Foundation Models · Tooling & Orchestration | 8/10 | Llama, PyTorch — the open-weight anchor. |
| SpaceX·xAI | 2002 | Foundation Models · Applications | 8/10 | Public markets priced a model lab inside a rocket company. Owns Cursor now. |
| Alibaba Qwen | 2023 | Foundation Models | 7/10 | Qwen3 72B+. Aggressively open-weight, fastest-closing competitor to US labs. |
| Databricks | 2013 | Cloud & Compute · Tooling & Orchestration · Foundation Models | 7/10 | Acquired MosaicML ($1.3B). Enterprise AI plumbing. |
| Waymo | 2009 | Applications · Foundation Models | 7/10 | The first physical AI at scale. 10M+ rides/month by 2025. |
| AI4Bharat | 2019 | Foundation Models | 6/10 | IIT Madras research. IndicNLP, IndicBERT, IndicTrans. The academic spine of Indic AI. |
| Apple Intelligence | 2024 | Foundation Models · Applications | 6/10 | On-device Foundation Models + Vision Pro. Privacy-first, slow, consumer-scale. |
| ByteDance Doubao | 2023 | Foundation Models · Applications | 6/10 | Doubao 1.5 Pro. TikTok's parent quietly shipped a frontier Chinese model. |
| DeepSeek | 2023 | Foundation Models | 6/10 | Shocked the market Jan 2025 with R1. MoE efficiency. |
| Midjourney | 2021 | Applications · Foundation Models | 6/10 | Image gen leader; profitable without VC. |
| Perplexity (co.) | 2022 | Applications · Foundation Models | 6/10 | Builds own search index + fine-tunes. Vertical from day one. |
| Physical Intelligence (π) | 2024 | Foundation Models · Applications | 6/10 | π₀ — general-purpose robot foundation model. Levine + Hausman + ex-Google Brain. |
| Sarvam AI | 2023 | Foundation Models | 6/10 | Indic-first foundation models. Sarvam-1, Sarvam-2B. Lightspeed + Khosla-backed. |
| xAI | 2023 | Foundation Models · Applications | 6/10 | Grok. Colossus supercluster in Memphis. |
| Baidu ERNIE | 2019 | Foundation Models · Applications | 5/10 | ERNIE 4.5 + Bot. Anchor of China's home-field LLM ecosystem. |
| ElevenLabs | 2022 | Applications · Foundation Models | 5/10 | Voice cloning + TTS. Owns audio. |
| Krutrim | 2023 | Foundation Models · Applications | 5/10 | Ola's AI lab. Open-sourced IndicLLM 7B/12B. Bhavish Aggarwal-backed. |
| Mistral | 2023 | Foundation Models | 5/10 | European frontier lab. Open-weight pragmatism. |
| Moonshot (Kimi) | 2023 | Foundation Models · Applications | 5/10 | Kimi k1.5 — China's first reasoning-model breakout. Million-token context. |
| NVIDIA GR00T · Isaac | 2024 | Tooling & Orchestration · Foundation Models | 5/10 | Humanoid foundation model platform + Isaac simulation. Robotics dev platform. |
| Runway | 2018 | Applications · Foundation Models | 5/10 | Gen-2/Gen-3 video models. Hollywood’s AI vendor. |
| Safe Superintelligence | 2024 | Foundation Models | 5/10 | Sutskever’s post-OpenAI lab. Zero products, $30B valuation. |
| Wayve | 2017 | Applications · Foundation Models | 5/10 | End-to-end learned driving + GAIA world model. UK AV champion. |
| World Labs | 2024 | Foundation Models | 5/10 | Fei-Fei Li's spatial-intelligence lab. $230M seed 2024, the highest-profile launch. |
| Cohere | 2019 | Foundation Models | 4/10 | Enterprise-focused. RAG-first positioning. |
| Figure | 2022 | Applications · Foundation Models | 4/10 | Humanoid robots. Ex-OpenAI Helix model. |
| MiniMax | 2021 | Foundation Models · Applications | 4/10 | MiniMax-01 with MoE attention. Hailuo video gen. |
| Skild AI | 2023 | Foundation Models | 4/10 | General-purpose embodied foundation model. $300M raised 2024. |
| Suno | 2022 | Applications · Foundation Models | 4/10 | Music generation. Current RIAA defendant. |
| Thinking Machines | 2024 | Foundation Models | 4/10 | Mira Murati’s lab. Post-OpenAI brain drain continues. |
| Zhipu AI | 2019 | Foundation Models | 4/10 | GLM-4. Tsinghua spin-off; hit by US entity list 2025. |
| AI21 Labs | 2017 | Foundation Models | 3/10 | Jamba, early GPT-3 competitor. |
| Inflection | 2022 | Foundation Models · Applications | 3/10 | Pi assistant. Team acqui-hired by Microsoft (2024). |
Tooling & Orchestration — 24 companies
| Company | Founded | Layers | Influence | Why it is here |
|---|---|---|---|---|
| OpenAI | 2015 | Foundation Models · Applications · Tooling & Orchestration | 10/10 | ChatGPT, GPT-4, Sora, o-series. Defined the category. |
| Anthropic | 2021 | Foundation Models · Applications · Tooling & Orchestration | 9/10 | Claude, MCP, constitutional AI. Safety-first frontier lab. |
| Hugging Face | 2016 | Tooling & Orchestration · Foundation Models | 8/10 | The GitHub of ML. 1M+ models hosted. |
| Meta AI | 2013 | Foundation Models · Tooling & Orchestration | 8/10 | Llama, PyTorch — the open-weight anchor. |
| Claude Code | 2025 | Applications · Tooling & Orchestration | 7/10 | Agentic CLI. Killer app for coding agents. |
| Databricks | 2013 | Cloud & Compute · Tooling & Orchestration · Foundation Models | 7/10 | Acquired MosaicML ($1.3B). Enterprise AI plumbing. |
| Vercel | 2015 | Tooling & Orchestration · Cloud & Compute | 7/10 | AI SDK, v0, AI Gateway. Frontend-first infra. |
| LangChain | 2022 | Tooling & Orchestration | 6/10 | Agent/chain orchestration — loved and loathed. |
| Qualcomm | 1985 | Silicon · Tooling & Orchestration | 6/10 | Bought Modular to fight CUDA with software. Silicon without a dev stack does not sell. |
| Scale AI | 2016 | Tooling & Orchestration | 6/10 | Data labeling → RLHF vendor. $870M Meta investment (2025). |
| CrewAI | 2023 | Tooling & Orchestration | 5/10 | Multi-agent framework. Role-playing agent crews. |
| LangGraph | 2024 | Tooling & Orchestration | 5/10 | Stateful graph-based agent orchestration from LangChain. |
| LlamaIndex | 2023 | Tooling & Orchestration | 5/10 | RAG-first data framework. |
| NVIDIA GR00T · Isaac | 2024 | Tooling & Orchestration · Foundation Models | 5/10 | Humanoid foundation model platform + Isaac simulation. Robotics dev platform. |
| Pinecone | 2019 | Tooling & Orchestration | 5/10 | Managed vector DB pioneer. |
| Replit | 2016 | Applications · Tooling & Orchestration | 5/10 | Agent IDE in the browser. Solo-founder-friendly. |
| Snowflake | 2012 | Cloud & Compute · Tooling & Orchestration | 5/10 | Cortex LLM functions. Enterprise data gravity. |
| Together AI | 2022 | Tooling & Orchestration · Cloud & Compute | 5/10 | Open-model inference cloud. |
| AutoGen | 2023 | Tooling & Orchestration | 4/10 | Microsoft Research multi-agent conversation framework. |
| Fireworks AI | 2022 | Tooling & Orchestration · Cloud & Compute | 4/10 | Low-latency inference for open models. |
| Replicate | 2019 | Tooling & Orchestration · Cloud & Compute | 4/10 | Run any model via HTTP. |
| Paperclip | 2025 | Tooling & Orchestration · Applications | 3/10 | Open-source AI labor management — agents with org charts, budgets, tickets. |
| Weaviate | 2019 | Tooling & Orchestration | 3/10 | Open-source vector database. |
| XR Blocks (Google) | 2026 | Tooling & Orchestration | 3/10 | WebXR + Gemini framework for AR prototyping. Spatial-AI on-ramp. |
Applications — 46 companies
| Company | Founded | Layers | Influence | Why it is here |
|---|---|---|---|---|
| ChatGPT | 2022 | Applications | 10/10 | The big bang. 100M users in 2 months. |
| Microsoft Azure | 2008 | Cloud & Compute · Foundation Models · Applications | 10/10 | OpenAI infra. Copilot everywhere. |
| OpenAI | 2015 | Foundation Models · Applications · Tooling & Orchestration | 10/10 | ChatGPT, GPT-4, Sora, o-series. Defined the category. |
| Anthropic | 2021 | Foundation Models · Applications · Tooling & Orchestration | 9/10 | Claude, MCP, constitutional AI. Safety-first frontier lab. |
| Google Cloud | 2008 | Cloud & Compute · Silicon · Foundation Models · Applications | 9/10 | TPUs + Gemini + Workspace. Only full-stack vertical. |
| Claude (app) | 2023 | Applications | 8/10 | Most-trusted assistant in developer and enterprise circles. |
| Cursor | 2022 | Applications | 8/10 | AI-native IDE. Fastest SaaS to $100M ARR. |
| GitHub Copilot | 2021 | Applications | 8/10 | First mass-market AI dev tool. |
| SpaceX·xAI | 2002 | Foundation Models · Applications | 8/10 | Public markets priced a model lab inside a rocket company. Owns Cursor now. |
| Claude Code | 2025 | Applications · Tooling & Orchestration | 7/10 | Agentic CLI. Killer app for coding agents. |
| Waymo | 2009 | Applications · Foundation Models | 7/10 | The first physical AI at scale. 10M+ rides/month by 2025. |
| Apple Intelligence | 2024 | Foundation Models · Applications | 6/10 | On-device Foundation Models + Vision Pro. Privacy-first, slow, consumer-scale. |
| Boston Dynamics | 1992 | Applications | 6/10 | Atlas, Spot, Stretch. The OG humanoid lab, now owned by Hyundai. |
| ByteDance Doubao | 2023 | Foundation Models · Applications | 6/10 | Doubao 1.5 Pro. TikTok's parent quietly shipped a frontier Chinese model. |
| Midjourney | 2021 | Applications · Foundation Models | 6/10 | Image gen leader; profitable without VC. |
| Perplexity | 2022 | Applications | 6/10 | Answer engine. Search reimagined. |
| Perplexity (co.) | 2022 | Applications · Foundation Models | 6/10 | Builds own search index + fine-tunes. Vertical from day one. |
| Physical Intelligence (π) | 2024 | Foundation Models · Applications | 6/10 | π₀ — general-purpose robot foundation model. Levine + Hausman + ex-Google Brain. |
| Salesforce | 1999 | Applications | 6/10 | Agentforce plus a $3.6B Fin acquisition — enterprise agents by M&A. |
| Tesla (Optimus, FSD) | 2003 | Applications · Silicon | 6/10 | Dojo chips + Optimus + FSD — an AI company hiding in a carmaker. |
| Unitree Robotics | 2016 | Applications | 6/10 | H1 + G1 humanoid at $16K. Quadrupeds at $1,600. China's price disruption. |
| xAI | 2023 | Foundation Models · Applications | 6/10 | Grok. Colossus supercluster in Memphis. |
| 1X Technologies | 2014 | Applications | 5/10 | Norwegian humanoid maker. NEO home robot. OpenAI-backed. |
| Apptronik | 2016 | Applications | 5/10 | Apollo humanoid shipping into Mercedes + GXO 2025. Texas-scale manufacturing bet. |
| Baidu ERNIE | 2019 | Foundation Models · Applications | 5/10 | ERNIE 4.5 + Bot. Anchor of China's home-field LLM ecosystem. |
| Bhashini | 2022 | Applications | 5/10 | Govt of India translation platform. National-scale public infra for 22 Indian languages. |
| Devin (Cognition) | 2023 | Applications | 5/10 | First "AI software engineer" framing. |
| ElevenLabs | 2022 | Applications · Foundation Models | 5/10 | Voice cloning + TTS. Owns audio. |
| Harvey | 2022 | Applications | 5/10 | Legal AI. Vertical-SaaS playbook for AI. |
| Krutrim | 2023 | Foundation Models · Applications | 5/10 | Ola's AI lab. Open-sourced IndicLLM 7B/12B. Bhavish Aggarwal-backed. |
| Moonshot (Kimi) | 2023 | Foundation Models · Applications | 5/10 | Kimi k1.5 — China's first reasoning-model breakout. Million-token context. |
| Replit | 2016 | Applications · Tooling & Orchestration | 5/10 | Agent IDE in the browser. Solo-founder-friendly. |
| Runway | 2018 | Applications · Foundation Models | 5/10 | Gen-2/Gen-3 video models. Hollywood’s AI vendor. |
| Wayve | 2017 | Applications · Foundation Models | 5/10 | End-to-end learned driving + GAIA world model. UK AV champion. |
| Yellow.ai | 2016 | Applications | 5/10 | Enterprise conversational AI. Customer-service bots at scale across 80+ countries. |
| Agility Robotics | 2015 | Applications | 4/10 | Digit — first humanoid deployed in GXO warehouses at paid production scale. |
| Character.AI | 2021 | Applications | 4/10 | Companion chatbots; acqui-hired by Google 2024. |
| Figure | 2022 | Applications · Foundation Models | 4/10 | Humanoid robots. Ex-OpenAI Helix model. |
| Fractal Analytics | 2000 | Applications | 4/10 | AI services + analytics. Predictive + generative across Fortune 500 enterprise. |
| Haptik | 2013 | Applications | 4/10 | Reliance Jio's chatbot platform. India's largest deployed conversational AI. |
| MiniMax | 2021 | Foundation Models · Applications | 4/10 | MiniMax-01 with MoE attention. Hailuo video gen. |
| Suno | 2022 | Applications · Foundation Models | 4/10 | Music generation. Current RIAA defendant. |
| Windsurf | 2021 | Applications | 4/10 | Cascade agent. Acquired by OpenAI (2025). |
| Inflection | 2022 | Foundation Models · Applications | 3/10 | Pi assistant. Team acqui-hired by Microsoft (2024). |
| Niramai | 2016 | Applications | 3/10 | AI thermal-imaging breast cancer screening. Healthcare AI from India. |
| Paperclip | 2025 | Tooling & Orchestration · Applications | 3/10 | Open-source AI labor management — agents with org charts, budgets, tickets. |
All 155 events, by era
Magnitude 3 is rare — fewer than 25 events across seventy years carry it. Anyone reusing this dataset should hold to the same standard.
Symbolic AI (1956–2011) — 15 events
Logic, expert systems, handcrafted features. Winters and thaws.
| Year | Event | Layer | Magnitude | What happened |
|---|---|---|---|---|
| 1956 | Dartmouth Workshop | Foundation Models | 3 | The term "artificial intelligence" is coined. |
| 1958 | Perceptron (Rosenblatt) | Foundation Models | 2 | First trainable neural network. |
| 1966 | ELIZA | Applications | 2 | Weizenbaum’s chatbot — first uncanny-valley moment. |
| 1968 | Intel founded | Silicon | 2 | The company that will miss the AI wave it enabled. (Intel) |
| 1986 | Backprop popularized | Foundation Models | 2 | Rumelhart, Hinton, Williams. Key algorithm returns. |
| 1987 | TSMC founded | Silicon | 3 | The silent kingmaker of every AI chip era. (TSMC) |
| 1993 | NVIDIA founded | Silicon | 3 | Gaming GPU company accidentally builds the AI stack. (NVIDIA) |
| 1997 | Deep Blue beats Kasparov | Applications | 2 | Symbolic AI’s last hurrah. |
| 1997 | LSTM (Hochreiter, Schmidhuber) | Foundation Models | 1 | Sequence modeling foundation for 15 years. |
| 2004 | DARPA Grand Challenge | Applications | 2 | No vehicle finishes. 2005 rematch — Stanley wins. Autonomous driving leaves the lab. |
| 2006 | CUDA released | Silicon | 3 | General-purpose GPU compute — the sleeper hit. (NVIDIA) |
| 2006 | AWS EC2 launches | Cloud & Compute | 3 | Cloud is born. Training becomes rentable. (AWS) |
| 2008 | Azure + Google Cloud emerge | Cloud & Compute | 2 | Hyperscaler trio takes shape. (Microsoft Azure) |
| 2009 | Google starts self-driving car project | Applications | 2 | The seed that becomes Waymo — 15 years of patient capital. (Waymo) |
| 2010 | DeepMind founded | Foundation Models | 2 | Hassabis + Legg bet on AGI-scale RL. (Google DeepMind) |
Deep Learning (2012–2016) — 11 events
AlexNet proves GPUs + data + depth wins. CNNs everywhere.
| Year | Event | Layer | Magnitude | What happened |
|---|---|---|---|---|
| 2012 | AlexNet crushes ImageNet | Foundation Models | 3 | GPU + CNN + data. The deep-learning era begins. |
| 2013 | Meta AI (FAIR) founded | Foundation Models | 1 | LeCun joins Facebook. Open research ethos. (Meta AI) |
| 2013 | Word2Vec | Foundation Models | 1 | Mikolov et al. — vector semantics. The prehistory of embeddings. |
| 2013 | Boston Dynamics Atlas debut | Applications | 1 | The first humanoid that made the internet watch. DARPA Robotics Challenge flagship. (Boston Dynamics) |
| 2014 | GANs (Goodfellow) | Foundation Models | 2 | Generative models take flight. |
| 2014 | Amazon Alexa + voice assistants | Applications | 1 | Mass consumer AI — mostly disappointing. |
| 2015 | OpenAI founded | Foundation Models | 3 | Non-profit mission to ensure AGI benefits all humanity. (OpenAI) |
| 2016 | Google TPU v1 | Silicon | 2 | First mainstream AI ASIC. (Google Cloud) |
| 2016 | AlphaGo beats Lee Sedol | Applications | 3 | "Move 37." Intuition in silicon. (Google DeepMind) |
| 2016 | TensorFlow + PyTorch | Tooling & Orchestration | 2 | The framework wars. PyTorch will win research, TF industry (briefly). |
| 2016 | Facebook + Microsoft Marea cable lands | Cloud & Compute | 1 | Transatlantic 160 Tbps. First time Big Tech co-owned sea infrastructure at scale. (Microsoft Azure) |
Transformer (2017–2021) — 14 events
Attention + scale. Pretraining becomes the playbook.
| Year | Event | Layer | Magnitude | What happened |
|---|---|---|---|---|
| 2017 | "Attention Is All You Need" | Foundation Models | 3 | Transformer paper. Rewrote the field. |
| 2018 | BERT + GPT-1 | Foundation Models | 2 | Pretrain-then-fine-tune becomes default. (OpenAI) |
| 2018 | AlphaFold 1 | Foundation Models | 1 | Structural biology becomes tractable. Nobel bait. (Google DeepMind) |
| 2019 | GPT-2 — "too dangerous to release" | Foundation Models | 2 | Scaling laws hint. Partial release controversy. (OpenAI) |
| 2019 | Microsoft invests $1B in OpenAI | Cloud & Compute | 3 | The deal that reshaped the industry. (Microsoft Azure) |
| 2019 | 2Africa consortium announced | Cloud & Compute | 1 | Meta + China Mobile + Orange + Vodafone ring Africa with 45,000 km of fiber. |
| 2020 | GPT-3 (175B) | Foundation Models | 3 | Few-shot learning. The first time the world noticed. (OpenAI) |
| 2020 | NVIDIA acquires Mellanox ($6.9B) | Silicon | 2 | Networking = new moat. InfiniBand fabrics the AI era. (NVIDIA) |
| 2020 | AlphaFold 2 cracks CASP14 | Foundation Models | 2 | "A solution to the protein folding problem." (Google DeepMind) |
| 2020 | Scaling Laws for Neural LMs | Foundation Models | 2 | Kaplan et al. — the paper that justified every subsequent billion. (OpenAI) |
| 2021 | Anthropic founded | Foundation Models | 2 | Amodei siblings + ex-OpenAI team. Safety-first. (Anthropic) |
| 2021 | GitHub Copilot launches | Applications | 3 | First killer AI app for a global profession. (GitHub Copilot) |
| 2021 | DALL·E + Midjourney | Applications | 2 | Generative images break containment. (Midjourney) |
| 2021 | Tesla AI Day — Optimus announced | Applications | 2 | Musk commits humanoid production. Widely mocked — turns out to be prescient. (Tesla (Optimus, FSD)) |
ChatGPT Era (2022–2024) — 49 events
Consumer breakout. Every enterprise suddenly has an AI budget.
| Year | Event | Layer | Magnitude | What happened |
|---|---|---|---|---|
| 2022 | ChatGPT launches (Nov) | Applications | 3 | The Big Bang. 1M users in 5 days, 100M in 2 months. (ChatGPT) |
| 2022 | Stable Diffusion open-sourced | Foundation Models | 2 | Open-weight image gen explodes creativity (and lawsuits). |
| 2022 | LangChain released | Tooling & Orchestration | 1 | Agent framework boom begins. (LangChain) |
| 2022 | Cursor founded | Applications | 2 | AI-native IDE. Will reach $100M ARR in under 2 years. (Cursor) |
| 2022 | JEPA (LeCun) | Foundation Models | 1 | Joint-Embedding Predictive Architecture. LeCun's quiet bet against pure transformers. (Meta AI) |
| 2023 | GPT-4 | Foundation Models | 3 | Genuinely useful across domains. Multimodal. (OpenAI) |
| 2023 | Claude 2 public release | Foundation Models | 2 | 100K context window — a new frontier. (Anthropic) |
| 2023 | Llama + Llama 2 open-weights | Foundation Models | 3 | Meta torpedoes the closed-model moat. (Meta AI) |
| 2023 | Gemini 1.0 | Foundation Models | 2 | Google finally ships. Multimodal from birth. (Google DeepMind) |
| 2023 | Vercel AI SDK | Tooling & Orchestration | 1 | Streaming-first JS/TS framework for AI apps. (Vercel) |
| 2023 | NVIDIA H100 ramps | Silicon | 3 | $30K a chip. Everyone is allocation-starved. (NVIDIA) |
| 2023 | xAI + Mistral founded | Foundation Models | 1 | Musk and ex-Meta French team enter the race. (xAI) |
| 2023 | Databricks acquires MosaicML ($1.3B) | Cloud & Compute | 2 | Enterprise data stacks get serious about training. (Databricks) |
| 2023 | MSFT extends OpenAI deal to $10B+ | Cloud & Compute | 2 | The partnership that reshapes cloud economics for a decade. (Microsoft Azure) |
| 2023 | Amazon invests $4B in Anthropic | Foundation Models | 2 | Google follows with $2B. Frontier labs become cloud-anchored duopolies. (AWS) |
| 2023 | AutoGen + CrewAI arrive | Tooling & Orchestration | 1 | Multi-agent frameworks proliferate. Agents start talking to each other. (CrewAI) |
| 2023 | Mamba / State-space models | Foundation Models | 1 | Gu + Dao — linear-time sequence models. The first credible transformer alternative. |
| 2023 | Qwen + Baidu ERNIE launch | Foundation Models | 1 | China's frontier-lab race accelerates. Qwen opens up aggressively. (Alibaba Qwen) |
| 2023 | Figure raises + commits to humanoid | Applications | 1 | Adcock raises $70M for a bet nobody else had conviction on. (Figure) |
| 2024 | Claude 3 + GPT-4o + Gemini 1.5 | Foundation Models | 3 | Capability parity at the frontier. Long context becomes normal. (Anthropic) |
| 2024 | NVIDIA Blackwell (B200) | Silicon | 2 | The H100 successor. Trillion-param models in reach. (NVIDIA) |
| 2024 | OpenAI o1 — reasoning chains | Foundation Models | 3 | Test-time compute = new scaling axis. (OpenAI) |
| 2024 | Anthropic ships MCP | Tooling & Orchestration | 2 | Model Context Protocol — USB-C for LLMs. (Anthropic) |
| 2024 | Devin framed as "AI engineer" | Applications | 1 | Agent hype peaks. Reality catches up slowly. (Devin (Cognition)) |
| 2024 | Stargate announced ($100B+) | Cloud & Compute | 2 | OpenAI + Oracle + SoftBank mega-cluster. (OpenAI) |
| 2024 | Character.AI acqui-hired by Google | Applications | 1 | First major frontier-era consolidation. (Character.AI) |
| 2024 | Sora (video gen) demoed | Applications | 2 | Minute-long coherent video. Hollywood notices. (OpenAI) |
| 2024 | Meta Llama 3 (405B open) | Foundation Models | 3 | Frontier-class weights released freely. Zuck doubles down. (Meta AI) |
| 2024 | Inflection team → Microsoft | Foundation Models | 1 | Regulatory workaround for the modern era. (Inflection) |
| 2024 | Ilya Sutskever launches SSI | Foundation Models | 2 | Raises $1B at $5B pre-revenue. Superintelligence as a product. (Safe Superintelligence) |
| 2024 | Claude computer use (beta) | Applications | 2 | Models see screens, click, type. Agents become ambient. (Anthropic) |
| 2024 | NVIDIA becomes $3T company | Silicon | 3 | Briefly the world’s most valuable company. Picks-and-shovels wins again. (NVIDIA) |
| 2024 | AlphaFold 3 + DeepMind Nobel | Applications | 2 | Hassabis + Jumper share chemistry Nobel. (Google DeepMind) |
| 2024 | xAI raises $6B at $24B | Foundation Models | 1 | Musk assembles 100K H100s in Memphis in months. (xAI) |
| 2024 | Cursor hits $100M ARR | Applications | 2 | Fastest SaaS in history — 12 months from zero. (Cursor) |
| 2024 | Perplexity $500M at $3B | Applications | 1 | Answer-engine category establishes itself. (Perplexity (co.)) |
| 2024 | LangGraph released | Tooling & Orchestration | 1 | Stateful graph agents — the infrastructure layer for long-running workflows. (LangGraph) |
| 2024 | Genie 1 (world model) | Applications | 2 | Foundation world model — generate interactive 2D environments from a single image. (Google DeepMind) |
| 2024 | V-JEPA 1.5 (self-supervised video) | Foundation Models | 1 | LeCun's roadmap to autonomous machine intelligence, one video frame at a time. (Meta AI) |
| 2024 | "Situational Awareness" published | Foundation Models | 2 | Leopold Aschenbrenner's essay reframes AGI as sovereign infrastructure. Shifts Capitol Hill. |
| 2024 | Physical Intelligence ships π₀ | Foundation Models | 3 | First generalist robot foundation model. VLA policies cross form factors — the GPT-3 moment for embodiment. (Physical Intelligence (π)) |
| 2024 | Figure 02 + OpenAI partnership | Applications | 2 | OpenAI models running on Figure humanoids. Natural-language-controlled manipulation. (Figure) |
| 2024 | Unitree G1 at $16K | Applications | 2 | Price floor collapses. China redefines humanoid economics overnight. (Unitree Robotics) |
| 2024 | NVIDIA Project GR00T | Tooling & Orchestration | 1 | Humanoid foundation-model platform + Isaac Sim. NVIDIA becomes the robotics dev cloud. (NVIDIA GR00T · Isaac) |
| 2024 | Waymo passes 1M rides/mo | Applications | 2 | Physical AI crosses the real-scale threshold. SF + Phoenix + LA + Austin operational. (Waymo) |
| 2024 | Microsoft + Constellation restart Three Mile Island | Cloud & Compute | 3 | The cold-war nuclear industry restarts for AI. Nuclear is suddenly cool again. (Constellation Energy) |
| 2024 | Amazon acquires Talen's Cumulus data center | Cloud & Compute | 2 | Nuclear-adjacent 960MW campus. Regulatory pushback follows — grid access becomes the new allocation fight. (AWS) |
| 2024 | Google signs for Kairos SMR 500MW | Cloud & Compute | 1 | First commercial SMR offtake in the AI era. Delivery 2030. (Kairos Power) |
| 2024 | Meta nuclear RFP | Cloud & Compute | 1 | Public call for 1–4GW of nuclear. The hyperscalers quietly price in 2030's grid. (Meta AI) |
Agentic (2025–2027) — 65 events
Models plan, use tools, and ship work autonomously.
| Year | Event | Layer | Magnitude | What happened |
|---|---|---|---|---|
| 2025 | DeepSeek R1 shocks the market (Jan) | Foundation Models | 3 | GPT-4-class reasoning, open weights, trained on a shoestring. NVIDIA drops 17% in a day. (DeepSeek) |
| 2025 | Claude 3.7 + Opus 4 | Foundation Models | 2 | Agentic coding capabilities become table stakes. (Anthropic) |
| 2025 | Claude Code + agentic CLIs explode | Applications | 3 | Terminal becomes the new IDE. Agents ship real PRs. (Claude Code) |
| 2025 | Groq + Cerebras land enterprise | Silicon | 1 | Inference-cost war starts. NVIDIA still training king. (Groq) |
| 2025 | Neocloud IPOs | Cloud & Compute | 2 | CoreWeave public. Neoclouds re-price data-center real estate. (CoreWeave) |
| 2025 | Cursor passes $500M ARR | Applications | 2 | Fastest SaaS in history. Windsurf acquired by OpenAI. (Cursor) |
| 2025 | Mira Murati → Thinking Machines | Foundation Models | 1 | OpenAI brain drain accelerates. (Thinking Machines) |
| 2025 | Windsurf acquired by OpenAI | Applications | 1 | Frontier labs start buying distribution. (Windsurf) |
| 2025 | Scale AI $870M Meta investment | Tooling & Orchestration | 1 | Data-labeling becomes strategic infrastructure. (Scale AI) |
| 2025 | Humanoid robots leave the lab | Applications | 2 | Figure, Tesla Optimus, 1X — embodiment era begins. (Figure) |
| 2025 | Stargate formally announced | Cloud & Compute | 2 | Up to $500B over 4 years. Sovereign-scale AI infra. (OpenAI) |
| 2025 | Anthropic $3.5B at $61.5B | Foundation Models | 2 | Valuation triples in 12 months. Enterprise Claude adoption compounds. (Anthropic) |
| 2025 | OpenAI $40B round ($300B+ valuation) | Foundation Models | 3 | Largest private funding round in history. SoftBank anchor. (OpenAI) |
| 2025 | Figure raises $675M at $2.6B | Applications | 1 | Humanoids are the new frontier bet. (Figure) |
| 2025 | Paperclip — agent labor mgmt | Tooling & Orchestration | 1 | MIT-licensed org-chart + budgets + tickets for AI workers. The HR layer for agents. (Paperclip) |
| 2025 | V-JEPA 2 + Genie 2 | Foundation Models | 1 | World-model architectures go multi-modal + controllable. Parallel track to LLM scaling. (Meta AI) |
| 2025 | Moonshot Kimi k1.5 reasoning breakout | Foundation Models | 1 | First Chinese reasoning-tier model. Million-token context beats US offerings. (Moonshot (Kimi)) |
| 2025 | Huawei Ascend 910C ships at scale | Silicon | 2 | US export controls backfire — China builds a domestic GPU supply chain in 3 years. (Huawei (HiSilicon)) |
| 2025 | Apptronik + Mercedes production | Applications | 2 | Humanoids enter production floors as paid labor. Not a pilot — a contract. (Apptronik) |
| 2025 | xAI Memphis Colossus — 150MW→1GW in 18 months | Cloud & Compute | 2 | Musk proves sovereign-scale clusters can be built faster than utilities thought possible. (xAI) |
| 2025 | Broadcom Tomahawk 6 / Arista 7800 AI Spine ship | Silicon | 1 | Ethernet-first AI fabric. NVIDIA Spectrum-X gets real competition. (Broadcom) |
| 2026 | Claude Opus 4.7 (1M ctx) | Foundation Models | 2 | Long-context reasoning becomes practical default. (Anthropic) |
| 2026 | Blackwell Ultra + Rubin previews | Silicon | 2 | Annual cadence locked in. Memory bandwidth is the new moat. (NVIDIA) |
| 2026 | MCP ecosystem 2,000+ servers | Tooling & Orchestration | 1 | De-facto standard for tool use across vendors. (Anthropic) |
| 2026 | Agents do real work (ops, eng, sales) | Applications | 2 | Multi-hour autonomous tasks land in production. |
| 2026 | Genie 3 mainstream · video world models | Foundation Models | 2 | Agents trained in generated environments. Sim-to-real pipeline becomes default. (Google DeepMind) |
| 2026 | ByteDance Doubao 1.5 Pro hits parity | Foundation Models | 1 | TikTok's parent closes the gap. Open-weight strategy erodes US closed moats. (ByteDance Doubao) |
| 2026 | Google XR Blocks v0.12 released | Tooling & Orchestration | 1 | WebXR + Gemini. Samsung Galaxy XR ships — spatial computing finally ships with AI native. (XR Blocks (Google)) |
| 2026 | Figure 03 at Tesla-scale manufacturing | Applications | 2 | Humanoids reach automotive-grade annual production volumes. (Figure) |
| 2026 | Washington switches off a frontier model | Foundation Models | 3 | Commerce export-control order forces Fable 5 and Mythos 5 dark worldwide — the first government recall of a commercial frontier model. (Anthropic) |
| 2026 | SpaceX swallows Cursor for $60B | Tooling & Orchestration | 2 | Largest startup acquisition ever, four days after SpaceX's own IPO. Every lab's biggest distribution channel now lives inside a rival. (Cursor) |
| 2026 | Claude Fable 5 + Mythos tier ships | Foundation Models | 2 | A Mythos-class model on general sale — for exactly three days. (Anthropic) |
| 2026 | GPT-5.6 launches to a government-approved list | Foundation Models | 2 | Twenty partners vetted by Washington; OpenAI says gating 'shouldn't become the default.' GA July 9. (OpenAI) |
| 2026 | Fable 5 returns; Sonnet 5 ships | Foundation Models | 2 | Order lifted after 18 days; Sonnet 5 lands near-Opus agentic performance at Sonnet prices. (Anthropic) |
| 2026 | SpaceX IPOs at $1.77T on an AI story | Foundation Models | 2 | Largest IPO ever ($75B raised); ~90% of the pitch was the xAI subsidiary. (SpaceX·xAI) |
| 2026 | Gemini 3.5 + Omni at I/O; Ultra cut to $200 | Foundation Models | 2 | Distribution beats margin — the price cut said the quiet part. (Google DeepMind) |
| 2026 | Apple ships Siri AI — with Gemini inside | Applications | 2 | Apple conceded the model layer to win back the assistant. (Apple Intelligence) |
| 2026 | Vera Rubin enters full production | Silicon | 2 | Blackwell's successor in full production; RTX Spark pitches agents-at-home. (NVIDIA) |
| 2026 | DeepSeek takes outside money at last | Foundation Models | 2 | $7.4B first external round; Tencent and CATL took lock-ups and zero votes — the state fund kept the voting rights. (DeepSeek) |
| 2026 | Qualcomm buys Modular to fight CUDA | Silicon | 2 | $3.9B for Mojo + MAX — buying the hardware-agnostic layer because silicon without a dev stack does not sell. (Qualcomm) |
| 2026 | Salesforce buys Fin for $3.6B | Applications | 2 | Paid $3.6B to bolt Intercom's ticket-resolving agent onto Agentforce rather than compete with it. (Salesforce) |
| 2026 | Grok 4.5, trained alongside Cursor | Foundation Models | 2 | First visible product of the Cursor deal, three weeks after closing. (xAI) |
| 2026 | EU softens the AI Act, keeps the GPAI hammer | Foundation Models | 1 | High-risk deadlines pushed 16 months; 3%-of-turnover GPAI fines still start August 2. |
| 2026 | NVIDIA books an $81.6B quarter | Silicon | 1 | Revenue up 85%, data center up 92% — the boom's most expensive routine earnings beat. (NVIDIA) |
| 2026 | Just Ask Jev: Reinforcement Learning for Calibrated Decisions as a Zero-Shot Detector of AI Alignment Failures | Foundation Models | 2 | Detectors of alignment failures screen deployed language models and score alignment benchmarks without task-specific fine-tuning. (Meta AI) |
| 2026 | ViRDM: Taming Representation Distribution Matching for Few-Step Causal Video Generation | Foundation Models | 2 | Few-step autoregressive video diffusion enables low-latency streaming generation using representation distribution matching. |
| 2026 | AgentKernel: The Trust-Native Agentic Operating System | Tooling & Orchestration | 1 | Architects explicit trust boundaries for autonomous agent runtimes handling privileged actions and untrusted web inputs. |
| 2026 | Agentic Detection of Online Conspiracies | Tooling & Orchestration | 1 | Multi-agent framework analyzes latent rhetorical structures and coordinated messaging patterns across conversational graphs. |
| 2026 | Your Transformer Can Hold Two Thoughts at Once: Evidence of Linear Superposition in LLMs | Silicon | 1 | While Large Language Models (LLMs) rely on highly non-linear components, in this work we demonstrate that they exhibit fundamental linear... |
| 2026 | ZooWork-ShopRanker: An Open, Preference-Aligned E-Commerce Reranker | Tooling & Orchestration | 1 | Open rerankers trained for general web retrieval transfer imperfectly to e-commerce, where ranking decisions depend not only on topical r... |
| 2026 | SLCA-GRPO: Resolving Cross-Segment Credit Misattribution in Tool-Calling RL | Silicon | 1 | Tool-calling agents produce heterogeneous outputs, interleaving structured tool invocations with user-facing natural language summaries. ... |
| 2026 | IndicBankBench: Evaluating Safety and Reliability of Language Model Assistants in Indian Retail Banking | Tooling & Orchestration | 1 | Banking assistants must use account-specific information to answer requests and, in many cases, take actions through tools. Evaluating on... |
| 2026 | TRACE: Temporal Audit and Condition-aware Evaluation of Streaming Video Understanding | Tooling & Orchestration | 1 | Streaming video understanding requires models to interpret evidence as it arrives, yet current evaluations often report task scores witho... |
| 2026 | Softmax Reparameterization for Output-Head Quantization | Silicon | 1 | Large vocabularies make output heads a substantial inference cost in small language models. We propose softmax reparameterization, a post... |
| 2026 | MOPD-Router: Rethinking Teacher Routing in Multi-Teacher On-Policy Distillation | Foundation Models | 2 | Multi-teacher on-policy distillation (MOPD) integrates specialized capabilities into a single student, but existing practice typically ha... |
| 2026 | Evidence-Grounded Auditing of Identification Assumptions in Climate-Policy Causal Evaluations | Tooling & Orchestration | 1 | Difference-in-differences (DID) studies are widely used to evaluate climate policy, but assessing the evidence supporting their identific... |
| 2026 | LightMIS: Ultra-Lightweight Medical Image Segmentation Without a Stage-Wise Decoder | Silicon | 1 | We present LightMIS, a scalable family of ultra-lightweight convolutional networks for 2D binary medical image segmentation without a lea... |
| 2026 | Not All Ranks Are Equal: Budget-Aware LoRA Merging Across Tasks | Foundation Models | 2 | Merging low-rank adapters (LoRAs) promises to eliminate the overhead of swapping task-specific weights at inference time. However, existi... |
| 2026 | Disaggregated Quantization: Specializing LLM Prefill and Decode | Foundation Models | 2 | Prefill and decode reward different approaches to quantization: low-precision arithmetic accelerates prompt processing, while compact wei... |
| 2026 | When Do Model Internals Help? Exploring the Role of Representation Engineering in LLM Safety | Silicon | 1 | Reliable AI safeguards require both control mechanisms that reduce unsafe behavior and monitoring mechanisms that detect safety risks dur... |
| 2026 | How Reproducible Are Evaluation Conclusions? A Self-Audit of LLM-Inferred Prompt Structure | Tooling & Orchestration | 1 | Evaluations of LLM systems routinely average over small prompt sets and report models as a ranked table. We ask how much confidence such ... |
| 2026 | WideSWE: Can Coding Agents Coordinate Changes Across Repositories? | Tooling & Orchestration | 1 | Coding-agent evaluation has progressed from resolving individual issues to carrying out long-horizon development, yet task completion is ... |
| 2026 | Groupwise Agentic Grading and Advantage Redistribution for Code Agent RL | Tooling & Orchestration | 1 | Reinforcement learning (RL) for code agents often uses executable tests to provide binary rewards. With these rewards, Group Relative Pol... |
| 2026 | Nereus: Adaptive Parallelism for LLM Post-Training | Silicon | 1 | Reinforcement learning (RL) post-training for large language models (LLMs) coordinates multiple models across generation, inference, and ... |
| 2026 | Recursive Harness Distillation across Agents for Robot Manipulation | Foundation Models | 2 | A central goal in robotics is to enable manipulation across changing tasks and environments. Vision-language-action (VLA) models provide ... |
33 people
People are first-class entities here, not attributes of a company, because in this industry they move and the companies follow.
| Name | Role | Now | Note |
|---|---|---|---|
| Geoffrey Hinton | Prof Emeritus · "Godfather of Deep Learning" · Nobel 2024 | U. Toronto | Backprop, RBMs, Capsule Nets. Walked away from Google in 2023 to speak on AI risk. |
| Yann LeCun | VP & Chief AI Scientist · Meta AI · NYU | Meta | CNNs, JEPA. Public champion of open weights + world-model architectures over pure LLM scaling. |
| Yoshua Bengio | Prof · Founder · Turing Award | Mila · Université de Montréal | Deep learning pioneer; now focused on AI safety and the LawZero initiative. |
| Ilya Sutskever | Co-founder · Safe Superintelligence | SSI | Ex-OpenAI co-founder. Launched SSI 2024 with zero products, $30B valuation by 2025. |
| Sam Altman | CEO | OpenAI | Architect of the $500B Stargate commitment. Defines the tempo of the frontier era. |
| Dario Amodei | CEO · Co-founder | Anthropic | Ex-OpenAI research lead. Author of "Machines of Loving Grace." Constitutional AI, Claude. |
| Daniela Amodei | President · Co-founder | Anthropic | Operating counterweight to Dario. Runs the actual company. |
| Demis Hassabis | CEO · Co-founder · Nobel 2024 | Google DeepMind | AlphaGo → AlphaFold → Gemini. Chemistry Nobel for protein structure prediction. |
| Jensen Huang | CEO · Co-founder | NVIDIA | Built the CUDA moat over 20 years. Briefly ran the world's most valuable company in 2024. |
| Mark Zuckerberg | CEO | Meta | Torpedoing closed-model moats with Llama + spending $60B/yr on compute. |
| Elon Musk | Founder · CEO | xAI / Tesla | Colossus in Memphis: 100K+ H100s standing in months. Grok as a real frontier competitor. |
| Sundar Pichai | CEO | Alphabet / Google | The only CEO running a fully vertically-integrated AI stack (chip → cloud → model → app). |
| Satya Nadella | CEO | Microsoft | The $13B OpenAI bet that reshaped the cloud era. Copilot everywhere. |
| Andrew Ng | Founder · Coursera co-founder | DeepLearning.AI · Landing AI | Taught ML to the world. RealAvatar.ai is his first AI-avatar experiment in 2024. |
| Fei-Fei Li | Co-founder · Stanford Prof | World Labs · Stanford HAI | ImageNet. Now building spatial intelligence at World Labs ($230M Series A, 2024). |
| Andrej Karpathy | Founder | Eureka Labs | Tesla → OpenAI → Eureka. "Zero to Hero," nanochat, autoresearch. The best teacher in AI. |
| Mira Murati | Founder · CEO | Thinking Machines | Ex-OpenAI CTO. Raised $2B seed in 2025 — the largest seed round in history. |
| Mustafa Suleyman | CEO | Microsoft AI | DeepMind co-founder → Inflection → Microsoft AI. Runs Copilot. |
| Liang Wenfeng | Founder | DeepSeek / High-Flyer | Ex-quant. Shipped GPT-4-class reasoning for <$6M. Broke the frontier-price assumption. |
| Jim Keller | CEO | Tenstorrent | Designed Apple A4, AMD Zen, Tesla FSD. Betting RISC-V + open silicon against NVIDIA. |
| Clem Delangue | CEO · Co-founder | Hugging Face | Turned a chatbot company into the GitHub of ML. 1M+ models, every frontier release. |
| Aravind Srinivas | CEO · Co-founder | Perplexity | Redefined search as an answer engine. $9B valuation by 2025. |
| Michael Truell | CEO · Co-founder | Anysphere (Cursor) | Took MIT → fastest SaaS to $500M ARR in history. 24-year-old founder generation. |
| Chris Olah | Co-founder · Interpretability | Anthropic | Built the field of mechanistic interpretability. The "how do these things work" school. |
| Leopold Aschenbrenner | Founder | Situational Awareness LLC | 2024 essay reframed AGI as sovereign infrastructure. Shifted Capitol Hill + Wall Street priors. |
| Noam Shazeer | Co-lead Gemini | Google DeepMind | Transformer co-author, MoE pioneer. Character.AI → reacquired by Google 2024. |
| Percy Liang | Prof · HELM lead | Stanford HAI | Runs the most trusted independent benchmark. The referee the labs actually cite. |
| Jeff Dean | Chief Scientist | TPUs, TensorFlow, AlphaFold infra. The backbone of Google's entire AI ops. | |
| Ashish Vaswani | Founder | Essential AI | First author of "Attention Is All You Need." Raised $56M for enterprise foundation models. |
| Brett Adcock | Founder · CEO | Figure | Serial founder (Archer Aviation). Shipped Figure 02 partnership with OpenAI in 2024. |
| Sergey Levine | Co-founder · Prof UC Berkeley | Physical Intelligence | Led π₀ — first generalist robot foundation model. The "GPT-3 moment" for embodiment. |
| Bernt Børnich | Founder · CEO | 1X Technologies | Norwegian-led humanoid startup. OpenAI-backed. Home robot (NEO) target consumer-first. |
| Rodney Brooks | Prof Emeritus · Founder | Robust.AI · MIT CSAIL | Behavior-based robotics pioneer. iRobot, Rethink Robotics. The field's conscience on hype. |
12 places to learn
| Programme | By | Kind | Note |
|---|---|---|---|
| DeepLearning.AI | Andrew Ng | courses | The canonical on-ramp. Short courses with every major lab. |
| NVIDIA Deep Learning Institute | NVIDIA | hands-on | GPU-accelerated workshops with certification. Most recognized credential in industry. |
| Hugging Face Learn | Hugging Face | courses | NLP, Deep RL, Audio, Diffusion. Always current because HF hosts the models. |
| OpenAI Academy | OpenAI | free | Launched 2025. Free AI literacy + builder tracks. Aimed at non-engineers. |
| Anthropic Courses | Anthropic | hands-on | Prompt engineering, tool use, real-world skills with Claude. Open in GitHub. |
| Google Machine Learning | courses | Crash Course still the best zero-to-intermediate ML foundation on the internet. | |
| Microsoft Learn · AI Engineer | Microsoft | certification | Azure + Copilot-focused learning path with AI-102 cert. |
| Meta AI Research | Meta | research | Primary sources for Llama, V-JEPA, Segment Anything, and DINO papers. |
| fast.ai | Jeremy Howard | free | Top-down practical deep learning. The antidote to math-first intros. |
| MIT 6.S191 · Intro to Deep Learning | MIT | free | Refreshed annually. Full video lectures + labs free online. |
| Stanford CS Courses | Stanford | academic | CS229 (ML), CS224N (NLP), CS236 (GenAI). Still the gold standard. |
| Vercel AI Documentation | Vercel | builder | Production-grade AI app patterns in TS. Streaming, tool use, multi-model. |
Reusing this
The dataset is CC BY 4.0. Cite it as: The Five Layer AI Cake — ConeApex (2026), https://coneapex.com · CC BY 4.0. The raw data is data.js; the six claims it was built to support are on the theses page.
The Five Layer AI Cake is published by ConeApex, the trading name of Nirukta Digital XR Technologies Private Limited (CIN U62090TS2025PTC198275), Hyderabad, Telangana, India. The dataset is licensed CC BY 4.0 — cite it freely. See llms.txt for how to attribute it.