Here's the original Narco video piped.video/watch?v=S73zDvMg…

Thrilled to share that @Qualcomm has officially completed its acquisition of @Modular 🚀 Together, we are building the leading unified AI compute platform for an array of high-growth areas, including data center, edge infrastructure, personal and industrial AI 🔥 We will continue to drive our an open platform to all hardware makers in the ecosystem with Mojo, MAX and Modular Cloud to enormous scale across CPUs, GPUs, NPUs, and custom silicon.

Today, @Qualcomm completed its acquisition of Modular. Our AI-native software platform pairs naturally with Qualcomm's leadership in high-performance, energy-efficient compute. Together, we're building a heterogeneous ecosystem open to all hardware makers, with leading performance across CPUs, GPUs, NPUs, and custom silicon. Our co-founder and CEO @clattner_llvm is taking on a new role as EVP of Advanced AI Software and Platforms, and our co-founder and president @iamtimdavis will be SVP and GM of Modular. The acquisition will accelerate the expansion of Qualcomm’ AI platforms and bring our technology to more developers, enterprises, hardware platforms, and markets. Full press release: modular.com/blog/qualcomm-co…

A pivotal moment for AI. As agentic workloads scale across data centers and edge, the industry needs a more open and modern software foundation. With @Modular, we are bringing developers an open AI-native software platform that runs across diverse compute environments and gives customers real choice in how and where they deploy AI.

We’re excited to announce that Modular has entered an agreement to be acquired by @Qualcomm. The future of unified compute has never been stronger. Read the full announcement: modular.com/blog/qualcomm-to…

.@iamtimdavis built a retro platformer where the terrain is generated by SIMD kernel computation. Meet GPU Boost Adventure. A neon, CRT-flavored endless runner. boost.modular.com

Gemma 4 is live on Modular Cloud, day zero, with the fastest performance on both NVIDIA and AMD. Our MAX inference framework delivers 15% higher throughput vs. vLLM on B200, and we’re the only inference provider to ship @googlegemma 4 on a framework we built ourselves. Two multimodal models live now: Gemma 4 31B (dense, 256K context) and 26B A4B (MoE, only 4B params active per pass). Both SOTA on Modular Cloud: modular.com/blog/day-zero-la… Modular Cloud runs on MAX, our inference framework that unifies GPU kernels, graph compilation, and high-performance serving in a single hardware-agnostic stack. New weights to SOTA deployment in days, on two hardware platforms: modular.com/request-demo?utm…

Disappointed to see this. Love @opencode, and I have huge respect for @AnthropicAI, but a big win for @OpenAI and an open developer ecosystem

MAX (AKA @iamtimdavis) is at Booth #3004 this week, suited up and ready to talk AI inference, hand you a Modular tote bag, and launch you back into orbit. No rocket science required. @NVIDIAGTC

🪫🔌🔋Here's a table I made that most folks don't fully appreciate in terms of AI and power. It maps NVIDIA GPU's from Ampere to Feynman and by all estimates in 2030, the most powerful NVIDIA GPU will draw 6kW - an order of magnitude more an A100, and the estimates are it'll deliver 250x the compute (at FP4/FP2). If we layer in what we know about frontier model scaling: ➡️ GPT-4 training: ~50 GWh ➡️ Gemini Ultra: ~50-75 GWh (estimated) ➡️ GPT-5+ class models: 500+ GWh (projected) ➡️ 2028 frontier models: 2-5 TWh (if scaling laws hold) At 2-5 TWh, you're talking about 0.5-1% of total US electricity generation for a single training run. 1,000 Feynman Ultra racks = 1 nuclear reactor running 24/7. This is why: ☢️ Microsoft is restarting Three Mile Island ⚡️ Amazon bought a nuclear-powered datacenter campus 🔋 Google signed the largest corporate PPA in history 🔌 Oracle is designing datacenters with small modular reactors (SMRs) on-site If it isn't already obvious - the companies that will train the next generation of frontier models are competing on gigawatts. Jensen wasn't being hyperbolic when he said "a gigawatt per rack sounds like a good limit." He was describing the inevitable. The question isn't whether AI will reshape our energy infrastructure. It's whether our infrastructure can scale fast enough to keep up with AI. Power is the new MOAT (and GOAT) of AI.