Who should I follow to learn about AI Infrastructure?

If by **AI Infrastructure** you mean the systems underneath modern AI - GPUs, clusters, networking, distributed training, inference, storage, Kubernetes, scheduling, and data centers - I’d follow a mix of **hands-on engineers, systems researchers, and infrastructure companies**, rather than generic “AI influencers.” ## My shortlist - **Amin Vahdat** - Google’s SVP/Chief Technologist for AI & Infrastructure. Excellent for understanding hyperscale networking, distributed systems, TPU infrastructure, and where AI infrastructure is going. Google’s 2026 material is particularly good on the shift toward infrastructure for agentic workloads. [AI infrastructure at Next ‘26 | Google Cloud Blog](https://cloud.google.com/blog/products/compute/ai-infrastructure-at-next26?utm_source=chatgpt.com) - **Ian Buck** - NVIDIA VP of Hyperscale & HPC. Follow him for GPU computing, hyperscale clusters, networking, and the architecture of modern AI factories. He is also giving the keynote at NVIDIA’s 2026 AI Infra Summit. [AI Infra Summit 2026 | NVIDIA Events](https://www.nvidia.com/en-us/events/ai-infra-summit/?utm_source=chatgpt.com) - **Anton Polyakov** - NVIDIA Director of Engineering working directly on AI supercomputers, Kubernetes, GPU hardware, cloud infrastructure and distributed systems. His technical posts are unusually close to the actual engineering. [Author: Anton Polyakov | NVIDIA Technical Blog](https://developer.nvidia.com/blog/author/apolyakov/?utm_source=chatgpt.com) - **Bill Dally** - NVIDIA Chief Scientist. Particularly valuable for the **hardware/architecture** side: GPUs, interconnects, memory, and accelerated computing. - **Jeff Dean** - Google. Essential if you want the foundational distributed-systems perspective behind Google's large-scale ML infrastructure. - **Matei Zaharia** - Stanford / Databricks. Great intersection of distributed systems, ML systems, serving, data infrastructure and developer tooling. - **Ion Stoica** - UC Berkeley / Anyscale. Follow for distributed systems and the infrastructure needed to make AI workloads practical at scale. - **Percy Liang** - Stanford. More model/systems oriented, but particularly useful for understanding the infrastructure implications of foundation-model development and evaluation. ### Companies/projects worth following - **NVIDIA** - probably the single most important source for understanding the current AI infrastructure stack. Their research team spans accelerated computing, systems and AI. [People | Research](https://research.nvidia.com/people?utm_source=chatgpt.com) - **Google Cloud / Google Research** - excellent for TPU architecture, distributed training and hyperscale infrastructure. - **CNCF / Kubernetes** - increasingly important because production AI is becoming a distributed-systems and orchestration problem. CNCF notes that Kubernetes is already widely used for generative-AI inference workloads. [The future of AI is community driven and open | CNCF](https://www.cncf.io/blog/2026/07/23/the-future-of-ai-is-community-driven-and-open/?utm_source=chatgpt.com) - **vLLM** - follow the project and its contributors if you're interested in **LLM inference**, serving, KV-cache management and GPU utilization. - **PyTorch** - particularly its distributed and compiler work. - **DeepSpeed** - excellent for large-scale training and inference systems. ## If you want to go *really* deep I'd organize your feed around six layers: 1. **Silicon:** GPUs, TPUs, accelerators, HBM 2. **Interconnect:** NVLink, InfiniBand, Ethernet, RDMA 3. **Distributed compute:** training, parallelism, schedulers, Kubernetes/Slurm 4. **Storage & data:** object storage, distributed filesystems, data pipelines 5. **Inference:** vLLM, batching, KV cache, quantization, serving 6. **Physical infrastructure:** data centers, power, cooling, networking and energy That last layer is becoming especially important: AI infrastructure is increasingly constrained not just by chips but by **power, data-center construction and time-to-power**. [AI Agents Are Thirsty for Power](https://www.wired.com/story/ai-agents-are-thirsty-for-power?utm_source=chatgpt.com) For newsletters, **Gradient Brief** is a good technical-oriented feed covering MLOps and AI infrastructure, while **RCRTech's AI Infrastructure Daily** is broader and more focused on the physical infrastructure/data-center/chip side. [Gradient Brief | MLOps & AI Infrastructure Newsletter | Gradient Brief](https://www.gradientbrief.com/?utm_source=chatgpt.com) **If I could give you only 10 follows:** Amin Vahdat, Ian Buck, Anton Polyakov, Bill Dally, Jeff Dean, Matei Zaharia, Ion Stoica, NVIDIA Research, vLLM, and CNCF/Kubernetes. If your goal is **to become an AI infrastructure engineer rather than just follow the industry**, I can also give you a **6-month AI Infra learning path - from Linux/networking → CUDA → distributed systems → Kubernetes → GPU clusters → LLM inference**, with the exact papers, repos, courses, and projects I'd do.

**If you want to learn about AI infrastructure, follow leaders who bridge hardware, cloud systems, and large-scale AI deployment - people like Jensen Huang (NVIDIA), Demis Hassabis (DeepMind), Dario Amodei (Anthropic), and Cassie Kozyrkov (decision intelligence). They provide insights into both the technical backbone and the organizational strategies behind modern AI.** ## 🔑 Key Figures in AI Infrastructure Name| Role| Why Follow Them| Platforms ---|---|---|--- **Jensen Huang**| CEO, NVIDIA| Architect of GPU-driven AI infrastructure; insights on hardware scaling, data centers, and AI-native computing.| X, LinkedIn [Top AI Leaders of 2026 | The Industry Leaders](https://www.theindustryleaders.org/leaders/top-25-ai-leaders-2026) **Demis Hassabis**| CEO, Google DeepMind| Leads multimodal AI and scientific applications; strong perspective on compute allocation and infrastructure for AGI.| X, LinkedIn [Top 10 AI Influencers of 2026 - KDnuggets](https://www.kdnuggets.com/top-10-ai-influencers-of-2026) **Dario Amodei**| CEO, Anthropic| Focuses on alignment and scaling infrastructure for Claude models; shares practical views on safe deployment.| X, LinkedIn [Top 50 People in AI to Follow in 2026 (X, LinkedIn, Websites)](https://www.aitraining2u.com/blog/top-50-ai-people-to-follow-2026.html) **Ilya Sutskever**| Co-founder, Safe Superintelligence| Advocates for massive compute-driven research; reshaping how labs allocate infrastructure resources.| Podcasts, Interviews [Top 10 AI Influencers of 2026 - KDnuggets](https://www.kdnuggets.com/top-10-ai-influencers-of-2026) **Cassie Kozyrkov**| Founder, Kozyr| Former Google Chief Decision Scientist; explains how organizations should structure data pipelines and AI infrastructure for decision-making.| LinkedIn, Blog [30 Top AI Thought Leaders To Watch](https://www.linkedin.com/pulse/30-top-ai-thought-leaders-watch-akhilesh-singh-pmp-caio-caiso-f9nxc) **Fei-Fei Li**| Director, Stanford HAI| Focuses on human-centered AI and infrastructure for large-scale datasets; bridges academia and enterprise.| LinkedIn, Stanford HAI [Top AI Leaders of 2026 | The Industry Leaders](https://www.theindustryleaders.org/leaders/top-25-ai-leaders-2026)[30 Top AI Thought Leaders To Watch](https://www.linkedin.com/pulse/30-top-ai-thought-leaders-watch-akhilesh-singh-pmp-caio-caiso-f9nxc) ## 📌 Why These Voices Matter * **Hardware & Compute Scaling**: Jensen Huang and Ilya Sutskever emphasize the role of GPUs, TPUs, and massive compute clusters in enabling frontier AI. * **Cloud & Deployment**: Dario Amodei and Cassie Kozyrkov provide practical insights into how organizations can build reliable, scalable AI infrastructure. * **Research & Ethics**: Demis Hassabis and Fei-Fei Li balance infrastructure discussions with ethical and scientific perspectives, ensuring infrastructure supports responsible AI. ## ⚠️ Risks & Trade-offs * **Compute Concentration**: Following only frontier lab leaders risks a narrow view - most emphasize scaling compute, which may not apply to smaller organizations. * **Bias Toward Big Tech**: Many voices are tied to large corporations; supplement with independent researchers (e.g., Timnit Gebru, Kate Crawford) for balance on ethics and infrastructure governance. [Top AI Leaders of 2026 | The Industry Leaders](https://www.theindustryleaders.org/leaders/top-25-ai-leaders-2026) * **Rapid Change**: AI infrastructure evolves quickly - cloud providers and hardware vendors release updates monthly. Following diverse voices ensures you don’t miss shifts in standards or best practices. 👉 To tailor your learning: * If you’re interested in **hardware and scaling**, focus on Jensen Huang and Ilya Sutskever. * For **enterprise deployment and decision-making**, Cassie Kozyrkov and Fei-Fei Li are essential. * For **research-driven infrastructure**, Demis Hassabis and Dario Amodei provide cutting-edge perspectives. Would you like me to curate a **shortlist of podcasts, newsletters, and blogs** these leaders contribute to, so you can follow them more efficiently?

To master **AI Infrastructure** - covering hardware like GPUs/TPUs, high-performance networking, distributed training, and cluster orchestration - you need to follow engineers, researchers, and organizations building the physical and software backbone of modern AI. [Google Cloud AI Infrastructure](https://www.skills.google/paths/2806#:~:text=AI%20Infrastructure%3A%20Cloud%20GPUs.%20Curious%20about%20the,makes%20each%20one%20unique%2C%20and%20how%20AI) Here are the top people, newsletters, and resources to follow: ### Industry Leaders & Engineers (X/Twitter & LinkedIn) * **Andrej Karpathy** (Former Director of AI at Tesla and co-founder of OpenAI) – While known for general AI, his technical breakdowns on compute, training runs, and software-hardware co-design are invaluable. * **Jeff Dean** (Chief Scientist at Google) – A pioneer in large-scale distributed systems and deep learning infrastructure (architect of MapReduce, TensorFlow, and TPU development). * **Soumith Chintala** (VP & AI Fellow at Meta, PyTorch creator) – Essential for tracking deep learning frameworks and hardware integration. * **Alexis Bjorlin** (VP at NVIDIA / former Intel/Google) – Deeply focused on scaling compute and cloud infrastructure for next-gen workloads. * **Speaks in Clusters / Systems Researchers** – Following computer architecture professors from institutions like MIT, Stanford, and UC Berkeley (such as **Tim Kraska** or **Matei Zaharia** for data systems) provides deep technical insight. ### Organizations & Institutional Resources * **NVIDIA Developer & AI Infrastructure Teams**: The absolute epicenter of hardware architecture. Follow their technical blogs and training paths (like the _NVIDIA-Certified Associate: AI Infrastructure and Operations_). [AI Infrastructure and Operations Fundamentals - Coursera](https://www.coursera.org/learn/ai-infrastructure-operations-fundamentals#:~:text=We%20start%20the%20journey%20with%20an%20Introduction,professional%20development.%20Who%20should%20take%20this%20course%3F) * **The Databricks Blog & Primer Systems**: Excellent for understanding the storage, data pipelines, and orchestration layers required before models even touch GPUs. * **Major Cloud Architecture Blogs (AWS, Google Cloud, Azure)** : Google Cloud's AI Infrastructure paths and AWS builders' libraries regularly publish deep dives into TPU/GPU cluster scaling, latency optimization, and InfiniBand networking. ### Newsletters & Communities * **The Gradient / Import AI (Jack Clark)** : Great for understanding how hardware limits shape AI capability and industry trends. * **Hugging Face Blog** : Excellent for infrastructure topics specifically related to efficient model serving, quantization, vLLM, and cluster inference. * **Semianalysis (Dylan Patel)** : The gold standard newsletter for deep-dive silicon supply chains, data center power constraints, advanced packaging, and AI infrastructure economics. What specific layer of AI infrastructure are you most interested in - the hardware/silicon level, cluster networking, or MLOps and deployment orchestration?

To truly understand AI Infrastructure - spanning large-scale GPU clusters, distributed training, low-latency inference, data centers, and the networking required for frontier models - you need to follow the engineers and researchers building the physical and software "plumbing" of the AI revolution. Here are the top voices, researchers, and engineers to follow, organized by their area of focus: 1\. The Deep Technical & Systems Educators * * * ** _Andrej Karpathy_** ([@karpathy on X](https://x.com/karpathy)): Former Director of AI at Tesla and co-founder of OpenAI. While he covers all of AI, his deep technical breakdowns, coding walkthroughs, and explanations of how neural networks actually execute on hardware are unmatched for infrastructure learners. * * * **_François Chollet_** ([@fchollet on X](https://x.com/fchollet)): Creator of Keras and developer of ARC-AGI. He frequently posts rigorous, foundational insights on the true mechanics of intelligence, software frameworks, and systems limitations. * * * **_Simon Willison_** ([@simonw on X](https://x.com/simonw)): Excellent follow for practical data engineering, local inference tooling, and how data pipelines intersect with modern AI applications. [15 AI Twitter Accounts to Follow for LLM Research and Insights](https://www.linkedin.com/posts/promptgenix_15-ai-related-accounts-you-should-follow-activity-7450520895883747328-8r0q)[The Best AI Engineers to Follow on X in 2026 | daily.dev](https://daily.dev/blog/best-ai-engineers-to-follow-on-x/)[20 Best AI Accounts to Follow on X (Twitter) in 2026 - Uxcel](https://uxcel.com/blog/best-ai-accounts-to-follow) 2\. The Scaling, Hardware & Architecture Debaters * * * ** _Yann LeCun_** ([@ylecun on X](https://x.com/ylecun)): Chief AI Scientist at Meta and Turing Award winner. Follow him for foundational hardware-to-algorithm perspectives and healthy, research-backed pushback on standard industry scaling narratives. * * * **_Demis Hassabis_** ([@demishassabis on X](https://x.com/demishassabis)): CEO of Google DeepMind. Essential for tracking how frontier scientific breakthroughs demand completely new paradigms in compute and lab-scale infrastructure. [daily.dev +2] 3\. Industry Infrastructure & Cloud Engineering Insiders * * To track massive-scale cluster operations, look to engineering leadership blogs and executive updates from the major compute providers driving the physical buildout: * **Meta AI Infrastructure / Engineering Teams** : Meta's massive open-source contributions to PyTorch scaling, custom silicon (MTIA), and data center networking offer some of the clearest public architectural deep dives. * **AWS & Microsoft Azure AI Blogs**: Following principal engineers from Microsoft Azure's AI Infrastructure team provides pragmatic views on globally distributed inference and multi-tenant training clusters. [2 AI Stocks To Buy Before 2027!!](https://www.youtube.com/watch?v=oUgnvamgNjo) * Would you prefer to focus more on the **software layer** (distributed training frameworks like PyTorch/Megatron, Ray, Kubernetes) or the **hardware/datacenter layer** (GPUs, networking fabrics like InfiniBand, power/cooling)? I can tailor specific whitepapers or repositories to look into.