Muse Spark Launches After Complete AI Stack Rebuild
1/ today we're releasing muse spark, the first model from MSL. nine months ago we rebuilt our ai stack from scratch. new infrastructure, new architecture, new data pipelines. muse spark is the result of that work, and now it powers meta ai. 🧵 https://t.co/fThDXdsxwB
Simplify Agent Architecture: Choose Workflow Over Multi-Agent
You don't need to overcomplicate your Agent Architecture. Do you also jump to multi-agent when a simple workflow would do the job faster, cheaper, and with far less debugging? I made a free Agent Architecture Cheatsheet to help you decide: - Workflow vs...
AQUA Rewritten in React: Faster, Smoother, Near Release
Our CTO @BraginRights and the @AquaBitcoin dev team have been working on a complete re-write of AQUA (Flutter to React) and we’re getting closer to release. There’ll probably be 1-2 more AQUA Flutter patches left. The React version is really...
Accepting Platform Lock‑In Can Be a Strategic Bet
I was working with a client this week who was wrestling with a decision I see a lot of teams avoid confronting honestly: whether to build on a third-party platform and accept the lock-in, or build something more flexible that...
Hub‑Spoke Architecture Enhances Lending Protocol Flexibility
I've seen this somewhere before 👀 Jokes aside, I do think hub/spoke architecture is the right direction for a lending protocol to capture wide range of existing and future use-cases and access more granular risk controls. Good call. https://t.co/slQ6zbyYnp
AI Multiplies Engineers, Leaving PMs Squeezed—Hire More
Narrative violation: Anthropic's Head of Growth says we'll need more PMs, not fewer. "While PMs and designers are getting leverage from AI, engineering is getting the most leverage right now. If you think about a default team with 5 engineers, 1...
Meta's Superintelligence Labs Assembles Hardware Team for New AI Devices
Meta Superintelligence Labs is quietly building a hardware team Veteran engineer Rui Xu, with experience at ByteDance, Xiaomi, and Dreamer, will lead the team It hints that Meta's Superintelligence Labs is exploring new AI device types beyond smart glasses. MSL chief Alexandr Wang...
Google Must Unify Services Into a Closed AI Ecosystem
All Google needs to do now is to connect all its services in a closed OpenClaw-like system with some memory and services as skills and they will win the AI race
Microsoft Pivots to Own AI Models, Challenging OpenAI
Microsoft is moving toward greater independence in AI. After revising its agreement with OpenAI, the company is now building its own frontier models to compete more directly, signaling a shift from partner to competitor. The strategy comes with trade-offs. Heavy investment in...
Apple Shifts AI Strategy Toward Privacy and Integration
Apple may have fallen behind in the first wave of AI, but the story is not over. Its privacy-first approach limited access to the data needed for rapid progress, while competitors moved faster. Now, Apple is recalibrating, including partnerships like using...
All‑in‑One Tesla Chip Fab Accelerates Development
My idea of a good time is working with amazing engineers to create incredible technology 🤩 The Tesla chip research fab will have all the machines needed to do logic, memory, packing & masks in one building for a lightning fast...
Moonshot's Kimi K2.5 Trained On
In his GTC presentation, Yang outlined how Moonshot’s latest flagship model, Kimi K2.5, was trained using Nvidia’s H800 graphics processing units (GPUs), a chip Nvidia customised to comply with US export controls. https://t.co/A2wLRFn2wG
Ruthlessly Shed Legacy Scaffolding to Unlock AI Agent Gains
One of the biggest lessons thus far in building AI agents is you have to be brutally unsentimental in your architecture. The models get better and better at handling things you previously built scaffolding for, you need to ruthlessly jettison...
Leading AI Integration Across Go, Dart, Flutter, Genkit
My remit changed and now I also lead our product and engineering for Go, Dart, Flutter, and Genkit. I like this post about principles our @dart_lang and @FlutterDev team are thinking about for thoughtfully bringing AI to the stack. https://t.co/7dhU7bw7Oj
Tech Founders and Execs Are Returning to Hands‑on Engineering
Founders + C-level folks who come from a software engineering background are all getting back to it. Never seen eng leaders get hands-on with a previous technology at so many different places.
Meta Appoints Instagram Exec to Lead Edits Team
Instagram's Edits team is working overtime this week: 👀 Scalable Pod revealed Meta exec Brett Westervelt would be the head of Edits https://scalablepod.com/p/exclusive-meta-taps-instagram-exec-to-lead-edits 👀 Matt Navarra shared an interview with Matt Simari (Director of Product Management at IG) to unpack the strategy...
AI Coding Agents Driving Unprecedented CTO Adoption
I've been selling software for 15 years now. I've never seen as much "market pull" before as I'm seeing now. For the AI sr software engineer, we've built. CTOs are truly mind blown. Few insights that actually surprise me: - The...
Ant Digital Shows Fintech Can Match Hyperscalers with Full‑Stack AI
Ant Digital Unveils Full-Stack AGI Architecture at IDC Question: Can fintech giants compete with hyperscalers on full-stack AI? Ant Digital positioning with complete vertical integration: → proprietary foundation models (Ling-1T trillion params) → finance-grade agent platform (Agentar-Fin-R1) → specialized infrastructure (ZOLOZ...
WorkUp Platform Redefines Asset Strategy, Says Quarterbac CTO
Engineering Readiness: How Paul Starr and the WorkUp Platform from Quarterbac are Redefining Asset Strategy Meet Paul Starr, CTO & Head of Software Engineering at Quarterbac - the thought leader & innovator who gave life to the WorkUp data platform.. https://t.co/34fOM3r9Zh
Starknet Powers High‑Compute Apps with New Architecture
Some apps need different tech to power them. It's not just about more throughput, it's about a different kind of architecture. For things like gaming, perps exchanges (maybe also AI), the ability to handle big computational transactions makes all the difference. Starknet is...
Rust Powers Ultra-Fast AI Automation for VC Workflow
I needed AI-powered automation that could actually keep up with my day. Existing tools were too slow, too error-prone. So I rewrote everything in Rust — 45+ APIs that Claude Code uses to run my VC workflow end-to-end. I'm presenting the architecture...
Tech Firms Weigh Buying AI Models vs Building Own Stack
I am happy to see a lot more "build vs buy" discussions happening across more tech companies. As in "should we buy leading models [like OpenAI, Anthropic] and be dependent on their pricing and reliability, or should we build+own our inference...
Starknet’s Proven ZK‑VM Safeguards Trillions in Value
Which ZK tech can you trust with your Dollars? Of recent, in the cross fire of the battle between Canton and Ethereum and Solana, the question of ZK safety has come up. As co-founder of @Zcash and @StarkWareLtd, both powered by...
CPU, Not GPU, Becomes AI’s New Bottleneck
Generative AI was all about GPU bursts. Agentic AI changes everything - it's continuous orchestration, coordination, and real-time decisions. That's why the CPU is becoming the new bottleneck. Arm just placed a massive bet with its first-ever production AGI CPU, claiming 2x...
Future Silicon Will Be Built for Agentic AI
From here on out, all silicon architecture designs will be designed with agentic AI in mind. Basically, all silicon becomes agentic workload-focused.
AI Augments Knowledge Workers, Preserving Human Agency
Marcus Fontoura, CTO for Microsoft Azure, author of HUMAN AGENCY IN A DIGITAL WORLD, on impact of AI on knowledge workers https://t.co/DFOk3yJikn
Apple's Privacy Focus Limits AI Ambitions, Favors On-Device Models
I’ve argued that Apple is constrained by what it can do with AI by its privacy narrative. Apple *has* trained foundation models, and they have a proprietary ~3BN LLM available on device via Foundation Models Framework. They also offer other...
AI Dev Productivity Shift: Laura Tacho Joins AWS
A big move happened in the AI + dev productivity space: Laura Tacho (@rhein_wein) DX —> AWS Now I’m paying attention to AWS and what they’ll do here 👀 (I’m also no longer an advisor at DX) https://t.co/Tr3Kzij6eR
AI Gateway Simplifies Token Metering for Every AI Factory
Every company will become an AI factory, wherein the unit of production is the token. Tokens, however, create usage tracking billing headaches that are v unlike SaaS. ▲ AI Gateway now solves the metering problem across models & providers with...
ZEC Powers Private AI Platform Silo’s Payment Layer
“Why we chose ZEC as the payment layer for Silo, our private AI platform”—https://t.co/UFPqM53FCq https://t.co/wxZxEWbopF
Own Ideas Get Stolen; Prioritize Payment Service Decoupling
You're in a cross-functional strategy meeting, exhausted from trying to get your architecture proposal taken seriously. The senior developer repeats your exact idea verbatim as if he just invented it. Peer: (Leans back, steepling his fingers) "I've been thinking, what we really...
Arm Unveils Its First Full‑scale Datacenter CPU
A deep dive into Arm’s first full-fledged datacenter CPU. $ARM $NVDA $TSMC $AMD $INTC https://t.co/2LhQgo6i3s
Make GitHub Core Agentic Infrastructure, Ditch Copilot
Here’s what I’d do if I was in charge of GitHub, in order: 1. Establish a North Star plan around being critical infrastructure for agentic code lifecycles and determine a set of ways to measure that. 2. Fire everyone who works on...
Arm Shifts to Own AI Chip, Redefining Value Capture
Arm is betting its future on its own AI chip. The company expects a major revenue boost as it moves beyond licensing designs into building more of its own technology stack. That shift could reshape how it captures value in the...
Stripe's In‑Ads Checkout Makes Shopify Optional
Uhhh is this the beginning of the inevitable breakup between Stripe and Shopify? This is a big deal. It’s also a bunch of my predictions coming to life at once. Here's what's happening: Stripe is powering a native checkout inside Facebook ads....
Explaining CPUs, GPUs, TPUs, NPUs and Qualcomm’s Role
There's been some talk of data center CPUs today. What's the role of CPUs vs GPUs vs TPUs vs NPUs - and where does @Qualcomm fit in? I had to try to do a plain-English answer. (shot and edited on...
Musk's Terafab Ambitions Face Major Feasibility Hurdles
Musk outlines terafab to deliver massive compute, yet production limits, resource shortages, and launch demands raise serious feasibility concerns. https://t.co/5gxDE0tf4d
Agentic CPUs Drive Datacenter Growth Amid Specific Workloads
In this report, we detail why, and what workloads specifically, are causing the datacenter CPU growth cycle. A new architecture emerges as agentic native CPUs find their place in the market. Report: Secret Agent CPU https://t.co/jjdrB3k1Bt
Arm Unveils Own CPU, Meta First Buyer, China Market Open
Bigly....Arm launches its own CPU, with Meta as first customer Arm can sell this to China...... https://t.co/AMt1kLNiV5
Arm Shifts From Silent Licensor to Power‑Efficient Brand
. @Arm spent decades as the quiet IP licensor behind everyone else's brand. Owning the agentic CPU narrative as a named vendor is a different game entirely. Power efficiency was always the moat. Now they get to say so out...
ADRs: Concise Decision Logs Documenting Architectural Evolution
NEW POST ADRs are short documents that capture a decision, structured into logs that reveal a history of architectural thinking for a product https://t.co/llbshdFbyf
Panther Lake Revives Intel on Laptops, Fulfilling G
Panther Lake is a real turning point for Intel. I've been on AMD since switching to Linux, but with 18A delivering Gelsinger's dream from 2021, it's a privilege to run blue again on the laptop side. And Intel working directly...
AI Codegen Hype Ignores Deployment, Observability, and Reliability
The chatter about generating code with AI tools feels stuck at the "basic" level of... well, codegen, plus (perhaps) reviews and testing. I hear close to little talk about the things that come right after generating code: deploying, canarying, o11y, SLOs,...
OpenAI Unifies ChatGPT, Codex Into Single AI Workspace
OpenAI is moving toward an all-in-one AI platform. The goal is to bring ChatGPT, Codex and other tools into a single environment and reduce fragmentation. Instead of switching between apps, AI becomes one continuous workspace. This is a shift from using AI...
ASML's New Boss Calls for Simpler, AI-Ready Reorg
ASML needs to become simpler. That, too, is innovation. Marco Pieters, ASML “We have become too sluggish,” says Marco Pieters, ASML’s new technical boss. He is leading a sweeping reorganization to prepare the Veldhoven-based high-tech giant for the AI era. https://t.co/0vG7jthH2I
Unified AI Architecture Bridges Knowing‑Doing Gap
AI is falling short because most organizations haven’t built the architecture to turn insight into action. In my latest @Forbes article, I explore why closing the gap between knowing and doing requires unified AI, data, workflows, and governance. This makes...
Inside Perspective Explains Uber’s Switch From HipChat
This is SO SUCH A GOOD example on why some things that do not make sense from the outside, make perfect sense from the inside. In 2016, Uber moved to its own chat system, off of HipChat (that had a disastrous...
Build End‑to‑End ML with AWS: Glue to SageMaker
A real AWS Data Science pipeline looks like this: Raw data → S3 ETL → AWS Glue Query → Athena Training → SageMaker Deployment → Endpoints Monitoring → CloudWatch Add streaming with Kinesis and orchestration with Step Functions, and you have a full production ML platform. This is...