Advancing AI Development

AI holds great potential for the semiconductor industry and will kick-start the next round of innovation for faster, cheaper and more energy-efficient computation – that was my message today at SPIE Advanced Lithography + Patterning. I discussed the potential and the challenges that AI holds for our industry.   The potential is clearly huge. AI is rapidly integrated into applications, and high-performance compute is expected to underpin growth towards $1 trillion of semiconductor sales by 2030. The challenges are around the computing needs of AI models and related energy consumption. The compute workload of training a leading AI model has increased 16x every 2 years in recent years – much faster than the increase in computing power delivered by Moore’s law, which is about 2x every 2 years. The energy needed to train a leading model has not grown so steeply but still rose 10x every 2 years. This computing need has been met by building supercomputers and massive data centers. If you extrapolate these trends, training a leading AI model would need the entire world-wide electricity supply in about 10 years. That’s clearly not realistic, so the trend has to break, by training algorithms becoming more efficient and by chips becoming more efficient. In other words, the needs of AI will stimulate immense innovation in chip design and manufacturing – and the potential value of AI to our society will put urgency and funding behind that drive. As a consequence, chip makers are pulling all levers to accelerate semiconductor scaling. This includes lithographic “2D” scaling: shrinking the dimensions of transistors to pack more into a square millimeter. It will also include “3D” integration, with innovations like backside power delivery, transistor designs like gate-all-around, as well as stacking chips in the package, where holistic lithography will play a critical role to deliver performance requirements. ASML will support these trends through a comprehensive, holistic lithography portfolio. Our 0.33 NA/0.55 NA EUV lithography systems allow chip makers to shrink dimensions at the lowest possible cost on their critical layers, while tightly matched and highly productive DUV systems will continue to reduce cost. More than ever, metrology and inspections tools – whose data is fed into lithography control solutions that keep the patterning process operating within tight specs to deliver the highest possible production yields – will be essential to deliver 2D scaling and 3D integration processes. 3D integration requires wafer-to-wafer bonding, and we have demonstrated the capability to map the stresses and distortions that bonding creates and to compensate for them, reducing overlay errors for post-bonding patterning by 10x or more.   It was a pleasure catching up with the industry’s lithography and patterning experts in San Jose. I’m excited to see our collective innovation power having a go at these challenges. Together, we will push technology forward.

🚨 Elon Musk and Nvidia CEO Jensen Huang are urging students to look beyond just learning how to code. As AI gets better at handling repetitive tasks, both believe the real advantage will come from understanding how the world works. Through physics and math. Jensen Huang recently said that if he were graduating today, he'd focus on physics. He explained that future AI systems will need to work with the physical world, not just digital spaces. This means knowing how things move, how forces interact, and how systems behave in real life. Elon Musk has echoed the same idea. When asked about useful skills for the future, he pointed to physics, backed by math. At Tesla and SpaceX, his thinking is rooted in solving problems from the ground up using core principles, not just following existing methods. They’re not saying coding is useless. It still matters. But the next big opportunities will go to people who understand the systems AI is meant to model, control, and improve. In simple terms, learn how the world really works. Study the tough stuff. Physics and math build the kind of thinking that machines can’t easily replace. ------- Do you agree?

𝗜𝗳 𝘆𝗼𝘂 𝗳𝗼𝗹𝗹𝗼𝘄 𝘁𝗵𝗲 𝗻𝗲𝘄𝘀, 𝘆𝗼𝘂’𝘃𝗲 𝗽𝗿𝗼𝗯𝗮𝗯𝗹𝘆 𝘀𝗲𝗲𝗻 𝗶𝘁 𝗮𝗹𝗹: 𝗔𝗜 𝗶𝘀 𝗯𝗼𝗼𝗺𝗶𝗻𝗴. 𝗔𝗜 𝗶𝘀 𝗼𝘃𝗲𝗿𝗵𝘆𝗽𝗲𝗱. 𝗔𝗜 𝘄𝗶𝗹𝗹 𝘀𝗮𝘃𝗲 𝘂𝘀. 𝗔𝗜 𝘄𝗶𝗹𝗹 𝗱𝗲𝘀𝘁𝗿𝗼𝘆 𝗷𝗼𝗯𝘀. The Stanford University AI Index 2025 cuts through all of it. Produced by the Institute for Human-Centered Artificial Intelligence, it’s one of the most respected and data-driven reports on the state of AI today. Over 400+ pages of concrete insights — from technical benchmarks and real-world adoption to policy shifts, economic impact, education, and public sentiment. 𝗧𝗵𝗲 2025 𝗲𝗱𝗶𝘁𝗶𝗼𝗻 𝗱𝗿𝗼𝗽𝗽𝗲𝗱 𝗹𝗮𝘀𝘁 𝘄𝗲𝗲𝗸. 𝗛𝗲𝗿𝗲 𝗮𝗿𝗲 12 𝗸𝗲𝘆 𝘁𝗮𝗸𝗲𝗮𝘄𝗮𝘆𝘀: 1. 𝗕𝗲𝗻𝗰𝗵𝗺𝗮𝗿𝗸𝘀 𝗮𝗿𝗲 𝗯𝗲𝗶𝗻𝗴 𝗰𝗿𝘂𝘀𝗵𝗲𝗱. ➝ AI performance on complex reasoning and programming tasks surged by up to 67 percentage points in just one year. 2. 𝗔𝗜 𝗶𝘀 𝗻𝗼 𝗹𝗼𝗻𝗴𝗲𝗿 𝘀𝘁𝘂𝗰𝗸 𝗶𝗻 𝘁𝗵𝗲 𝗹𝗮𝗯. ➝ 223 FDA-approved AI medical devices. Over 150,000 autonomous rides weekly from Waymo. This is mainstream adoption. 3. 𝗕𝘂𝘀𝗶𝗻𝗲𝘀𝘀 𝗶𝘀 𝗴𝗼𝗶𝗻𝗴 𝗮𝗹𝗹-𝗶𝗻.  ➝ $109B in U.S. private AI investment. 78% of organizations using AI. Productivity gains are no longer theoretical. 4. 𝗧𝗵𝗲 𝗨.𝗦. 𝗹𝗲𝗮𝗱𝘀 𝗶𝗻 𝗾𝘂𝗮𝗻𝘁𝗶𝘁𝘆—𝗖𝗵𝗶𝗻𝗮’𝘀 𝗰𝗮𝘁𝗰𝗵𝗶𝗻𝗴 𝘂𝗽 𝗼𝗻 𝗾𝘂𝗮𝗹𝗶𝘁𝘆.  ➝ Chinese models now rival U.S. models on MMLU, HumanEval, and more. Global AI is becoming a multi-polar game. 5. 𝗥𝗲𝘀𝗽𝗼𝗻𝘀𝗶𝗯𝗹𝗲 𝗔𝗜 𝗶𝘀 𝗹𝗮𝗴𝗴𝗶𝗻𝗴 𝗯𝗲𝗵𝗶𝗻𝗱 𝗶𝗻𝗻𝗼𝘃𝗮𝘁𝗶𝗼𝗻. ➝ Incidents are rising, but standardized RAI benchmarks and audits are still rare.   Governments are stepping in faster than vendors. 6. 𝗚𝗹𝗼𝗯𝗮𝗹 𝗼𝗽𝘁𝗶𝗺𝗶𝘀𝗺 𝗶𝘀 𝗿𝗶𝘀𝗶𝗻𝗴—𝗯𝘂𝘁 𝗻𝗼𝘁 𝗲𝘃𝗲𝗻𝗹𝘆.   ➝ 83% of people in China are optimistic about AI. In the U.S., that number is just 39%. 7. 𝗔𝗜 𝗶𝘀 𝗴𝗲𝘁𝘁𝗶𝗻𝗴 𝗰𝗵𝗲𝗮𝗽𝗲𝗿, 𝘀𝗺𝗮𝗹𝗹𝗲𝗿, 𝗮𝗻𝗱 𝗳𝗮𝘀𝘁𝗲𝗿.  ➝ The cost of GPT-3.5-level inference dropped 280x in two years. Open-weight models are nearly matching closed ones. 8. 𝗚𝗼𝘃𝗲𝗿𝗻𝗺𝗲𝗻𝘁𝘀 𝗮𝗿𝗲 𝗿𝗲𝗴𝘂𝗹𝗮𝘁𝗶𝗻𝗴 𝗮𝗻𝗱 𝗶𝗻𝘃𝗲𝘀𝘁𝗶𝗻𝗴.  ➝ From Canada’s $2.4B to Saudi Arabia’s $100B push—states aren’t watching from the sidelines anymore. 9. 𝗘𝗱𝘂𝗰𝗮𝘁𝗶𝗼𝗻 𝗶𝘀 𝗲𝘅𝗽𝗮𝗻𝗱𝗶𝗻𝗴—𝗯𝘂𝘁 𝗿𝗲𝗮𝗱𝗶𝗻𝗲𝘀𝘀 𝗹𝗮𝗴𝘀. ➝ Access is improving, but infrastructure gaps and lack of teacher training still limit global reach. 10. 𝗜𝗻𝗱𝘂𝘀𝘁𝗿𝘆 𝗶𝘀 𝗱𝗼𝗺𝗶𝗻𝗮𝘁𝗶𝗻𝗴 𝗺𝗼𝗱𝗲𝗹 𝗱𝗲𝘃𝗲𝗹𝗼𝗽𝗺𝗲𝗻𝘁.   ➝ 90% of top AI models now come from companies—not academia. The gap between top players is shrinking fast. 11. 𝗔𝗜 𝗶𝘀 𝘀𝗵𝗮𝗽𝗶𝗻𝗴 𝘀𝗰𝗶𝗲𝗻𝗰𝗲.   ➝ AI-driven breakthroughs in physics, chemistry, and biology are earning Nobel Prizes and Turing Awards. 12. 𝗖𝗼𝗺𝗽𝗹𝗲𝘅 𝗿𝗲𝗮𝘀𝗼𝗻𝗶𝗻𝗴 𝗿𝗲𝗺𝗮𝗶𝗻𝘀 𝘁𝗵𝗲 𝗰𝗲𝗶𝗹𝗶𝗻𝗴.   ➝ Despite all the progress, models still struggle with logic-heavy tasks. Precision is still a challenge. You can download the full report FREE here: https://lnkd.in/dzzuE5tN

From CES this week, one thing is clear: we are moving into the era of physical AI — intelligence that operates in the real world.    Robotics, including humanoid and non-humanoid systems, are getting a lot of attention right now. This is familiar territory for Autodesk. We have decades of experience working with manufacturing, AI, and industrial design leaders who build in the physical world.    MarketWatch recently explored this momentum and included some of my perspective: https://lnkd.in/e_DN9HwC    Progress will not come from machines that just look like us, nor just language. It will come from AI that understands physics, objects, and three-dimensional space. That’s why work on world models, like what Fei-Fei Li and others are doing, matters. These systems learn from sensory data to build a usable understanding of their environment.    Physical AI will change how every industry that makes things designs, simulates, and executes. That is core to Autodesk’s mission, and I am optimistic about what is ahead.    Who is ready to put physical AI to work across everything we design and build? 

Claude Code's source code leaked last week. 512,000 lines of TypeScript. Most people focused on the drama. I focused on the memory architecture. Here's how Claude Code actually remembers things across sessions — and why it's a masterclass in agent design: 𝗧𝗵𝗲 𝟯-𝗟𝗮𝘆𝗲𝗿 𝗠𝗲𝗺𝗼𝗿𝘆 𝗔𝗿𝗰𝗵𝗶𝘁𝗲𝗰𝘁𝘂𝗿𝗲: 𝗟𝗮𝘆𝗲𝗿 𝟭 — 𝗠𝗘𝗠𝗢𝗥𝗬. 𝗺𝗱 (𝗔𝗹𝘄𝗮𝘆𝘀 𝗟𝗼𝗮𝗱𝗲𝗱) A lightweight index file. Not storage — pointers. Each line is under 150 characters. First 200 lines get injected into context at every session start. It points to topic files. It never holds the actual knowledge. Think of it as a table of contents, not the book. 𝗟𝗮𝘆𝗲𝗿 𝟮 — 𝗧𝗼𝗽𝗶𝗰 𝗙𝗶𝗹𝗲𝘀 (𝗢𝗻-𝗗𝗲𝗺𝗮𝗻𝗱) Detailed knowledge spread across separate markdown files. Architecture decisions. Naming conventions. Test commands. Loaded only when MEMORY. md says they're relevant. Not everything gets loaded. Only what's needed right now. 𝗟𝗮𝘆𝗲𝗿 𝟯 — 𝗥𝗮𝘄 𝗧𝗿𝗮𝗻𝘀𝗰𝗿𝗶𝗽𝘁𝘀 (𝗚𝗿𝗲𝗽-𝗕𝗮𝘀𝗲𝗱 𝗦𝗲𝗮𝗿𝗰𝗵) Past session transcripts are never fully reloaded. They're searched using grep for specific identifiers. Fast. Deterministic. No embeddings. No vector DB. Just plain text search when the first two layers aren't enough. But here's the part that blew my mind: 𝗦𝗸𝗲𝗽𝘁𝗶𝗰𝗮𝗹 𝗠𝗲𝗺𝗼𝗿𝘆. The agent treats its own memory as a hint, not a fact. Memory says a function exists? → Verify against the codebase first. Memory says a file is at this path? → Check before using it. And one more design principle hidden in the code: If something can be re-derived from source code — it doesn't get stored. Code patterns, conventions, architecture? Excluded from memory saves entirely. Because if it can be looked up, it shouldn't be remembered. 𝗪𝗵𝘆 𝘁𝗵𝗶𝘀 𝗺𝗮𝘁𝘁𝗲𝗿𝘀 𝗯𝗲𝘆𝗼𝗻𝗱 𝗖𝗹𝗮𝘂𝗱𝗲 𝗖𝗼𝗱𝗲: This 3-layer pattern is model-agnostic. Any team building AI agents can steal this: → Keep your always-loaded context tiny → Reference everything else via pointers → Never persist what can be looked up → Treat memory as a hint, not truth The future of AI agents isn't about how much they remember. It's about how well they forget. What memory patterns are you using in your agent builds?

A company I know deployed an AI agent in 3 days. No boundaries defined. No guardrails. No sandbox testing. No failure playbook. Week 1: It sent 400 unapproved emails to clients. This is not a horror story. This is what happens when excitement outpaces engineering. The companies succeeding with AI agents in 2026 all follow the same principle: Scaling follows confidence, not excitement. They start small. They define limits. They test adversarial scenarios. They build human approval gates. They observe before they expand. Here’s the step-by-step deployment path serious teams follow - Start with a safe, low-risk use case - Define the agent’s boundaries clearly - Map structured workflows (no guessing) - Ground it with trusted data sources - Apply least-privilege access - Add guardrails before autonomy - Choose the right architecture - Test in simulation (normal + edge cases) - Deploy in a sandbox first - Introduce human approval gates - Add observability and monitoring - Roll out gradually - Create a failure playbook - Build continuous learning loops - Implement governance & compliance controls Safe AI isn’t about slowing down innovation. It’s about engineering trust. Constrain → Ground → Test → Observe → Expand. 15-step framework. Swipe through. Your team needs this before the next sprint planning meeting. What’s the biggest mistake you’ve seen in AI agent deployment? Drop it below 👇

My biggest takeaways from Fei-Fei Li: 1. Just nine years ago, calling yourself an AI company was considered bad for business. Nobody believed the technology would work back in 2016. By 2017, companies started embracing the term. Today, virtually every company calls itself an AI company. 2. The modern AI revolution started with a simple but overlooked insight from Fei Fei: AI models needed large amounts of labeled data. While researchers focused on sophisticated mathematical models and algorithms, she realized the missing ingredient was data. Her team spent three years working with tens of thousands of people across more than 100 countries to label 15 million images, creating ImageNet. This dataset became the foundation for today’s AI systems. 3. The human brain’s efficiency vastly exceeds current AI systems. Humans operate on about 20 watts of power—less than any lightbulb—yet accomplish tasks that require AI systems to use massive computing resources. Current AI still can’t do things elementary school children find easy. 4. Simply scaling current approaches won’t be enough. While adding more data, computing power, and bigger models will continue advancing AI, fundamental innovations are still needed. Throughout AI history, simpler approaches combined with enormous datasets consistently outperformed sophisticated algorithms with limited data. 5. Breakthrough technologies often start as toys or fun experiments before changing the world. ChatGPT was tweeted by Sam Altman as “Here’s a cool thing we’re playing with” and became the fastest-growing product in history. What seems like play today might transform civilization tomorrow. 6. Spatial intelligence is as crucial as language for real-world applications. In emergency situations like fires or natural disasters, first responders organize rescue efforts through spatial awareness, movement coordination, and understanding physical environments—not primarily through language. This is why world models that understand three-dimensional space represent the next frontier beyond text-based chatbots. 7. Physical robots face much harder challenges than self-driving cars, which took 20 years from prototype to street deployment and still aren’t finished. Self-driving cars are metal boxes moving on flat surfaces, trying not to touch anything. Robots are three-dimensional objects moving in three-dimensional spaces, specifically trying to touch and manipulate things. This makes robotics far harder than creating chatbots. 8. Everyone has a role in AI’s future, regardless of profession. Whether you’re an artist using AI tools to tell unique stories, a farmer participating in community decisions about AI deployment, or a nurse who could benefit from AI assistance in an overworked health-care system, you can and should engage with this technology. AI should augment human dignity and agency, not replace it—which means both using AI as a tool and having a voice in how it’s governed.

If you are building AI agents or learning about them, then you should keep these best practices in mind 👇 Building agentic systems isn’t just about chaining prompts anymore, it’s about designing robust, interpretable, and production-grade systems that interact with tools, humans, and other agents in complex environments. Here are 10 essential design principles you need to know: ➡️ Modular Architectures Separate planning, reasoning, perception, and actuation. This makes your agents more interpretable and easier to debug. Think planner-executor separation in LangGraph or CogAgent-style designs. ➡️ Tool-Use APIs via MCP or Open Function Calling Adopt the Model Context Protocol (MCP) or OpenAI’s Function Calling to interface safely with external tools. These standard interfaces provide strong typing, parameter validation, and consistent execution behavior. ➡️ Long-Term & Working Memory Memory is non-optional for non-trivial agents. Use hybrid memory stacks, vector search tools like MemGPT or Marqo for retrieval, combined with structured memory systems like LlamaIndex agents for factual consistency. ➡️ Reflection & Self-Critique Loops Implement agent self-evaluation using ReAct, Reflexion, or emerging techniques like Voyager-style curriculum refinement. Reflection improves reasoning and helps correct hallucinated chains of thought. ➡️ Planning with Hierarchies Use hierarchical planning: a high-level planner for task decomposition and a low-level executor to interact with tools. This improves reusability and modularity, especially in multi-step or multi-modal workflows. ➡️ Multi-Agent Collaboration Use protocols like AutoGen, A2A, or ChatDev to support agent-to-agent negotiation, subtask allocation, and cooperative planning. This is foundational for open-ended workflows and enterprise-scale orchestration. ➡️ Simulation + Eval Harnesses Always test in simulation. Use benchmarks like ToolBench, SWE-agent, or AgentBoard to validate agent performance before production. This minimizes surprises and surfaces regressions early. ➡️ Safety & Alignment Layers Don’t ship agents without guardrails. Use tools like Llama Guard v4, Prompt Shield, and role-based access controls. Add structured rate-limiting to prevent overuse or sensitive tool invocation. ➡️ Cost-Aware Agent Execution Implement token budgeting, step count tracking, and execution metrics. Especially in multi-agent settings, costs can grow exponentially if unbounded. ➡️ Human-in-the-Loop Orchestration Always have an escalation path. Add override triggers, fallback LLMs, or route to human-in-the-loop for edge cases and critical decision points. This protects quality and trust. PS: If you are interested to learn more about AI Agents and MCP, join the hands-on workshop, I am hosting on 31st May: https://lnkd.in/dWyiN89z If you found this insightful, share this with your network ♻️ Follow me (Aishwarya Srinivasan) for more AI insights and educational content.

The AI race will not be won by the biggest model. It will be won by the most adaptable infrastructure. Three shifts stand out right now. 1. Image models are becoming reasoning engines.   Multimodal models are now generating accurate charts, legible text, and consistent layouts from natural language. Images are becoming a valid front end for analytics, training, and operations, not just brand and marketing. 2. Coding agents are moving from autocomplete to coworkers.   New models are built to work for hours on a problem, refactor large code bases, and manage complex workflows. This is the start of continuous software delivery by AI, not just a side tool for developers. 3. The AI infrastructure cycle is accelerating and becoming heterogeneous.   Inference is always on and needs to sit closer to data, users, and regulators. That is driving a build out of specialized compute across CPUs, GPUs, TPUs, and other accelerators. Frontier models like Gemini are already trained and served on custom TPUs, while GPUs remain the workhorses for parallel math and CPUs still anchor control and business logic. The question is no longer which chip to choose, but how to compose the right mix and move workloads as cost, regulation, and model options evolve. In my role at Rackspace I see this weekly with leaders in healthcare, financial services, and the public sector. They are not asking whether to use AI. They are asking how to secure the right mix of compute and locations without recreating technical debt. For forward deployed leaders, the ones closest to customers and operations, the agenda for the next 12 to 24 months is clear: • Treat image models as a new experience layer. Take one important customer or employee journey and redesign it so dynamic visuals and copilots are the primary interface, not static reports or dashboards.   • Select one critical workflow and rebuild it with AI at the center. Break it into steps, decide where agents own the work and where humans stay in the loop, and redesign the data and process around that.   • Plan capacity and partnerships around persistent inference demand and a mix of CPU, GPU, and TPU, rather than a single vendor or architecture. The gap will not be who has access to AI. It will be which organizations can rewire their operating model and infrastructure fast enough, while staying flexible enough to pivot as the landscape continues to shift.

I keep hearing studies about how women are behind in AI, and I can’t help but wonder if this is just sewing a biased seed. Like a self-fulfilling prophecy, tell women they’re behind and they will question whether they should start or continue going. We’re seeing the same narrative pattern as when we talk about women in cybersecurity—where women were the OG computers, programmers, cryptologists until men saw dollar signs and drove them out. Women have played pivotal founding roles in AI and, yet now we’re trying to convince them they’re late to the party because there’s money being made. If women are behind in AI, it’s because men are hogging the time to learn through disproportionate distribution of the mental load in families and organizations. Studies have proven that men have more leisure time than women, giving them an advantage of capacity to read, learn, and participate in conversations about AI. Bottom line is we need diverse perspectives building and interacting with AI to ensure mitigation of biased implementation and outputs. In order to make AI participation more inclusive, we need to address the systems and societal norms that are contributing to the imbalance of opportunity and signals of discouragement. Companies can achieve this through… … caregiver support … flexible work schedules … providing time on the clock for study … encouraging men to take leave for family … approve professional development requests for AI training equitably And also pay equitable wages to ensure women have equitable financial opportunity to pay for training out of pocket if that’s necessary. Let’s stop talking about women being behind in AI, and keep the conversation on actionable, equitable access and inclusion for everyone.