ICYMI this week at @WeAreDevs World Congress North America, the Arm team took the stage to explore how developers can build, deploy, and run AI directly on physical devices.
From navigating the practical challenges of constrained compute and connectivity to enabling on-device vision AI without relying on the cloud, the sessions offered a look at what it takes to bring AI into the physical world. 🚀
What’s slowing down your mobile game? 📱
Using Arm Performance Studio, we uncovered hidden GPU bottlenecks and cut frame time from 50ms to 25ms, with the game looking almost identical.
See how to find and fix geometry, overdraw and post-processing issues on real hardware. okt.to/vPfkcW
🏆 The results are in.
330+ submissions. Five winning projects across Cloud AI, Mobile AI, and Physical AI.
Congratulations to the winners of the Arm Create: AI Optimization Challenge and thank you to every developer who took part.
Meet the winners ➡️okt.to/Mljmi2
The Arm Create Builder Zone coming to @WeAreDevs World Congress North America!
🚀 Lightning talks
🚀 Demos
🚀 Conversations with the teams building the next era of AI
This is your chance to learn more about how YOU can build, test, and experiment on Arm.
Come and see us at booth 601: okt.to/oClepa
Is your Windows app ready for the Arm PC moment?
Join Arm, @Microsoft and @MaxonZbrush for our upcoming webinar to learn more about why Maxon went native with ZBrush, what users gain and how AppReady can help your team do the same.
📍 10:00 – 11:00 AM PT | October 1
🔗 Register now: okt.to/vgCM3c
AI is changing not only what developers build, but how they build it.
At Arm Create, Shantu Roy shared how the Arm AI Portal builds on decades of software investment to simplify the complexity of AI development and meet developers and agents where they build.
For more than 22 million developers building on Arm, that means less time navigating the complexity of AI development — and more time focused on what differentiates their applications.
🚀 Explore the Arm AI Portal now: okt.to/W1Zbyw
Your badge. Your code. Make it your own. 👾
At Arm Create, developers didn’t just wear their badges — they hacked them.
Personalize the screen, play Snake or Flappy Bird, track the International Space Station, or build something completely your own.
What would you make yours do?
Arm Create brought developers together in Shanghai and Shenzhen to build, experiment and get hands-on with AI.
Want to dig into what we explored? Watch Shantu Roy break down four key decisions you need to make to build production-ready AI systems. 👇
okt.to/rQl1h5
500+ developers came together today at Arm Create in Shanghai, getting hands-on with edge AI, cloud AI and physical AI.
Technical sessions. Live demos. Plenty being built.
Here’s a look at what they’ve been building. 👇
Arm AI Portal is here! 🎉
A new destination designed for you and your agents to discover and evaluate pre-optimized AI models, performance data, code examples and deployment workflows all in one place.
By bringing together optimized AI resources from across the ecosystem, you spend less time searching and benchmarking and more time building on Arm. okt.to/7sOBuy
Building custom silicon for increasingly demanding AI infrastructure means balancing performance, efficiency and differentiation - and doing it all without adding unnecessary integration complexity.
That's why we're excited to introduce Arm Neoverse CSS N4: Arm’s most configurable CSS yet.
That means more flexibility to differentiate. A faster path from CSS to silicon. ⚡️ okt.to/Pb5MI3
CSS for Mobile 2 is here. 📱
Built on Arm’s latest AI CPU and AI-native GPU technologies, it’s developer-ready from day one, bringing together KleidiAI, the Arm Neural Graphics Development Kit and open interfaces on a common platform to build and deploy AI at scale.
See what's new: okt.to/6UM9Oq
Two cities. One place to get hands-on with what’s next on Arm. 💻
We’ll be in Shanghai and Shenzhen next week for Arm Create, bringing together developers, engineers and ecosystem partners for technical talks, live demos and hands-on sessions across gaming, cloud, edge and robotics.
Follow us to stay close on the latest from Arm Create. okt.to/qWtI4b
👋 Say hello to @IBMcloud's first dual-architecture mainframe processor built to support operating systems and applications on both IBM and Arm architectures in future IBM Z and LinuxONE systems.
With more than 22 million software developers worldwide, the Arm software ecosystem is bringing its broad and growing range of applications to IBM platforms, while Arm workloads benefit from IBM’s enterprise-grade capabilities, including hardware-level fault detection and recovery, advanced encryption, secure key management, and AI acceleration.
This is the future of CPU innovation built on Arm in action.
okt.to/b8e04u
Wherever your AI runs, there's a place for it in the Arm AI Optimization Challenge.
Explore the Physical AI, Cloud AI, and Mobile AI tracks, build a project that demonstrates measurable optimization on Arm-powered platforms, and compete for a share of the $8,000 cash prize pool.
⏰Submissions close 14 August. okt.to/5wNF7c
Dynamic Insights in Arm Performix are here, bringing AI-guided performance analysis to agentic development.
➡️ Runtime performance data
➡️ Arm-developed analysis guidance
➡️ LLM reasoning
Delivering evidence-based recommendations that help uncover bottlenecks, identify root causes, and optimize software on Arm-based infrastructure.
okt.to/DExHdn
Migrating to Arm is only the beginning.
Together with @AWS_Partners, @Atlassian demonstrates how developers can optimize their workloads on Arm Neoverse-based AWS Graviton through production-scale testing, JVM tuning, and workload-specific optimization.
The results are higher performance, lower latency, and greater efficiency at scale. 🚀
okt.to/AmtTXU
That’s a wrap on #SIGGRAPH2026! 🙌
From Neural Dawn to Arm Model Gym and Vulkan workflows, we showed how developers can start building with neural graphics on Arm today.
Thanks to everyone who joined our demos, sessions and conversations.
Agentic AI gets more powerful when it works where developers already do. 👨💻
In just under a year, the Arm MCP Server has reached 10,000+ @Docker downloads, with developers integrating it across Amazon Kiro, Claude Code, Codex, Cursor, Gemini, GitHub Copilot, Google Antigravity, Windsurf and more.
Here's a closer look at the future of agentic AI development on Arm: okt.to/2OGQU4
Neural graphics are changing how every frame is built. 📱
Neural Dawn combines conventional rendering with Arm Neural Technology to deliver ray-traced lighting and increase the displayed frame rate from 30 to 60 FPS within a mobile power envelope.
Recently showcased at Arm Create during #SIGGRAPH2026.
Looking for ideas for your Arm AI Optimization Challenge submission? 💡
We've highlighted a few examples across Physical AI, Cloud AI, and Edge AI to help you think about what you could build, optimize, benchmark, or improve.
Projects demonstrating clear optimization work and measurable results will compete for a share of the $8,000 cash prize pool.
⏰Submissions close 14 August. okt.to/hwSY2j
That's a wrap on Arm Create at SIGGRAPH 2026! 🎬
Thank you to everyone who joined us and to our partners for helping showcase the future of mobile gaming.
From behind-the-scenes looks at Neural Dawn to live demos, technical sessions, and great conversations, it's always inspiring to see developers exploring what's possible on Arm.
There are only 4 days until Arm Create at #SIGGRAPH2026.
Register today to experience Neural Dawn through demos and expert conversations and learn how to start building with these technologies.
🔗: okt.to/LVM0eC
Neural Graphics on Arm is ready for production now.
Join us at Arm Create at #SIGGRAPH2026 to experience Neural Dawn firsthand through demos and conversations with the experts behind the tech, and learn how to start building with these technologies today.
👾: okt.to/ld0u31
Have an idea for the Arm AI Optimization Challenge? 👨💻
Join our office hours on Discord to ask questions, explore project ideas, and get guidance on building and optimizing AI for Arm-powered platforms.
Whether you're submitting to Cloud AI, Edge AI, or Physical AI, this is a great opportunity to get started. okt.to/7GhWSB
What could you build with a chance to win a share of $8,000?
The Arm AI Optimization Challenge is calling on developers to create and optimize AI workloads for Arm-powered platforms across Cloud AI, Edge AI, and Physical AI.
Show your optimization work, benchmark your results, and compete alongside developers from around the world.
📅 Entries close 14 August. okt.to/1t8UwL
Together with Sumo Digital, we're showcasing what's possible with Arm Neural Technology through Neural Dawn, a project to deliver a production-quality game. 🤝
Hear from Gary Dunn, Co-CEO and COO at Sumo Digital, on how our collaboration is helping make @UnrealEngine MegaLights, ray tracing and advanced lighting practical on future Arm CSS for mobile. okt.to/Kkfehp
Applications for the next Arm Ambassador cohort are now open. 🎉
If you're building, mentoring, or helping developers succeed on Arm, you were made for this. Join a community of technical leaders and experts making a real impact.
Apply today! ➡️ okt.to/O9YH38
Deploying AI models at the edge comes with a different set of challenges.
These hands-on Jupyter labs walk you through using ExecuTorch to deploy and optimize @PyTorch models on Arm CPUs and NPUs, with examples you can run on hardware including Raspberry Pi. okt.to/8Zhg3E
What does engineering intelligence look like in practice?
It's the creativity behind a breakthrough, the expertise behind a solution and the curiosity that keeps innovation moving forward.
As part of #INWED26, we took a moment with our partners @AstonMartinF1 to celebrate this year's theme, Engineering Intelligence, and recognize the women whose ideas, expertise and innovation help shape our industry every day.
This is your chance to build a local AI smart camera with @Raspberry_Pi 📸
Join our hands-on workshop and learn how to work with live camera input, run object detection locally, and see how to build an edge AI application that runs on Raspberry Pi.
Happening tomorrow! 📆 okt.to/KY4rEd
This is your chance to build a local AI smart camera with @Raspberry_Pi 📸
Join our hands-on workshop and learn how to work with live camera input, run object detection locally, and see how to build an edge AI application that runs on Raspberry Pi.
Happening tomorrow! 📆 okt.to/zr8yOv
If you build or run workloads on any Arm Neoverse platform, this session is for you.
Don't miss our upcoming office hours and join the team who built Arm Performix for a live demo and open Q&A where we will :
➡️ Profile a running workload and collect hardware-level performance data
➡️ Read guided insights across CPU, memory, and core utilization
➡️ Cut root cause analysis from hours to minutes
➡️ Integrate with your favorite LLM in agentic workflows
Tuesday, June 23 at 9 AM P.T | 17:00 B.S.T
American Sign Language to text. Real-time. On-device.
This open-source project demonstrates ASL recognition on Arm Ethos-U65 NPU, recognizing 29 characters with ~3.5ms inference and local processing for low-latency performance.
okt.to/o51EcZ
Calling all developers interested in Kubernetes deployment environments, Infrastructure as Code (IaC), @OctopusDeploy, and CI/CD automation! 📣
Don't miss this hands-on workshop demonstrating how to build a modern deployment pipeline for Arm-based workloads using Octopus Deploy, Terraform, and Amazon EKS. okt.to/Wokbix
Arm AppReady for Windows is here! 🎉
A new program bringing you AI-enabled workflows, technical guidance and access to app optimization experts all in one place, helping developers reduce porting complexity and build native Windows on Arm applications.
okt.to/qZg3fQ
Today at #MicrosoftBuild, @Microsoft announced the preview of Azure Cobalt 200 VMs.
Based on Microsoft agent developer benchmarks Cobalt 200 delivers:
⚡️16% faster agentic sandbox creation
⚡️42% higher agentic sandbox memory bandwidth
⚡️50% better agentic sandbox tool execution performance
Compared to the previous generation - delivering faster responsiveness, better utilization across cloud fleets and lower overall cost for running large-scale AI services. okt.to/Hgdwsa
Getting Python packages running natively on Windows on Arm isn't just about rebuilding code. Dependencies, build systems, and wheel availability can all get in the way.
A WoA porting agent helps automate the process, making it easier to uncover and address ecosystem gaps. 🧠
okt.to/wMHxaK
We’re empowering developers to build once, optimize efficiently and scale AI workloads, from cloud to edge. With a global ecosystem of 22M+ developers, we provide the platform and tools needed to unlock performance, efficiency and real-world impact.
At #ArmEverywhere, Sharbani Roy shared success in this new era of AI depends on three things: breadth, depth and value. okt.to/yDUmwL
Developers are building the future of AI and today we’re introducing new ways to help them do it. Say hello to:
🔹 Arm Create - hands-on events, sessions, and challenges
🔹 Arm Developer Program - tools, learning, community, and recognition
🔹 Kits for developers - practical assets and workflows to help developers start building on Arm
Together, these experiences are designed to make Arm easier to engage with, more practical to build on, and more connected across tools, community, and live participation.
Explore the new developer experience with Arm: okt.to/6eVRD1
Available as open source, Arm Metis is designed to identify complex vulnerabilities earlier across large-scale codebases with up to 10x higher true positive rates and ~50% fewer false positives.
As AI scales, the next generation of secure computing is evolving to keep up. 💪 okt.to/UD0Rdm
@Nvidia developers 🤝 Arm developers
We had a great time bringing our communities together for an evening of connection, knowledge, and inspiration. Thank you to all who joined us as we explored where the state of AI is now, and where it is headed next. 💡
What does it take to stay ahead in game development? 🎮
Join us for a Discord session with @EnduringGames to explore how early access to emerging technologies, from SDKs and plugins to engine integrations and new hardware features can help shape better gaming experiences.
We’ll cover:
➡️Benefits of early integration with partners like Arm and Unreal Engine
➡️Real-world development insights from Enduring Games
➡️Trade-offs and considerations at different stages of development
okt.to/4c16yV
Want to learn how to identify code hotspots within minutes, interpret performance data to prioritize optimization efforts, and move from raw metrics to concrete improvements?
Then don't miss next week's Code-Along for a closer look at how Arm Performix can be integrated into automated and agentic workflows!
➡️: okt.to/ErDGJ3
On-device AI shouldn’t mean slower performance or complex optimization.
With @GoogleAI Edge + Arm SME2, Stable Audio runs fully on-device — with 2x faster audio generation and 4x lower memory, no code changes required.
Start experimenting now using Google AI Edge tools and KleidiAI-accelerated LiteRT: okt.to/px9J4L
Join us for a developer meetup with @nvidia at our San Jose office on May 19 for an evening focused on AI compute, optimization and production-ready AI applications. 👨💻
Hear from Arm and NVIDIA experts on AI workloads, latency, quantization, frameworks, and performance optimization, plus live demos, networking, and a raffle with prizes for attendees. okt.to/frAWRi
RAG pipelines are only as strong as the infrastructure behind them. See how Arm Performix helps optimize AI-driven retrieval and generation on Arm Neoverse platforms, improving BLAS performance, throughput, and efficiency across modern cloud workloads. 💡
okt.to/kbnWd5
Image classification pipelines don’t always need bigger models to improve. 💡
Vociply used Arm Perfomix on AWS Graviton to get 40% faster performance and reduce cost - demonstrating how tuning on Arm can improve both speed and efficiency. okt.to/jzn7Qe
What developers can do better with Arm? Build, create, optimize, scale from model to edge - and so.much.more.
Thank you to everyone who joined us at our first ever Arm Create, we can't wait to continue building the future, together, on Arm! 🎉
Arm Performix - a new class of performance tooling built for agentic AI systems- is here! 🎉
Surfacing system-wide bottlenecks and turning low-level hardware data into actionable insights across agentic AI workloads: okt.to/N7e06l
We had such a great time connecting with all of you last week at #GoogleCloudNext and we can't wait to continue building the next generation of Google Cloud, on Arm, together. 🤝
What does it actually take to build intelligent systems that work reliably in the real world?
Join Arm experts and Ambassadors for a practical discussion on edge AI, embedded system design, and deploying real-time intelligence on on 28 April at 16:00 GMT.
From autonomous driving to deep learning at the edge, this session focuses on what engineers need to build production-ready systems today. okt.to/QXAidn
We're headed to Google Cloud Next!
We can't wait to connect with the developers driving the shift to Arm based infrastructure. For those attending, be sure to visit the Arm booth for demos and conversations about the future of cloud native and AI workloads running on Google Axion, built on Arm.
Join us in San Francisco for Arm Create, a hands-on event for developers working across AI, ML, and compute platforms, taking place alongside the AI Developer Conference by @DeepLearningAI.👨💻
Explore how AI models run across cloud and edge, build experiences in minutes, and connect with Arm engineers and the wider developer community, and enjoy drinks and bites throughout the evening. okt.to/FX7yxj