Moor Insights & Strategy

💡AI tokens are getting cheaper. Enterprise AI bills aren’t. As reasoning models generate more tokens and agentic AI multiplies the work behind every request, enterprises are facing a new infrastructure question: What does it cost to produce a useful AI outcome? In new research commissioned by Solidigm, Moor Insights & Strategy Vice President and Principal Analyst Matthew Kimball examines why storage is becoming an increasingly important part of that equation. The research brief explores: ⚙️ Why falling token prices don’t necessarily translate into lower AI costs ⚙️ How agentic AI changes the economics of inference ⚙️ The connection between storage performance and GPU utilization ⚙️ Why context reuse and KV-cache strategies matter ⚙️ How power, density and NVMe SSD architecture affect AI infrastructure economics ⚙️ Why storage should be considered alongside accelerators and networking as a day-zero architecture decision ➡️ Read the full brief: https://lnkd.in/gY9MYMQE Solidigm Patrick Moorhead

Just a little over 2 hours before Marc Benioff, CEO at Salesforce kicks off Day 1 of The Six Five Media Summit: AI Unleashed 2026👏🏻🎥 Followed by three days of conversations with technology leaders including: 🔹 Sridhar Ramaswamy, CEO, Snowflake 🔹 Matt Murphy, Chairman and CEO, Marvell Technology 🔹Gary Dickerson, President and CEO, Applied Materials Join us August 25–27, with streaming beginning TODAY at 10:00 a.m. CT🚨 📍Register here: https://lnkd.in/gKdvbHDj

Day 1 of the Six Five Media Summit: AI Unleashed 2026 starts TOMORROW 👏🏻🚨 Salesforce Chair & CEO Marc Benioff kicks things off, followed by Sridhar Ramaswm CEO of Snowflake headlining Day 1, Matt Murphy CEO of Marvell Technology opening Day 2, and Gary Dickerson CEO of Applied Materials bringing things home on Day 3. Each day will feature tracks built around a different pressure point: infrastructure, the edge, and the silicon underneath it all.💡 🎟️Register for the free Summit here: https://lnkd.in/e-rSR9Nt

CEO & Chief Analyst, Patrick Moorhead sat down with Modular, a Qualcomm company CEO Chris Lattner, and Qualcomm CEO Cristiano R. Amon on why the industry needs a better software model for heterogeneous compute and how Qualcomm + Modular intend to attack it💡 🎥View the full conversation here:

What’s ahead in tech this week? 👀 The AI bill is climbing—and the story is running on three clocks: chips coming, deployments landing, and earnings paying for both. ⏱️ In this week’s The Week Ahead in Technology, CEO & Chief Analyst, Patrick Moorhead, breaks down what he’s watching across the industry: 🔹 Chips: Hot Chips at Stanford brings one of the densest silicon disclosures of the year, with NVIDIA, AMD, Intel, Arm, Google, Microsoft, Meta, OpenAI and more presenting. The big takeaway? Custom silicon is increasingly sitting alongside the GPU—not replacing it. 🔹 Deployments: Patrick is co-hosting The Six Five Media Summit: AI Unleashed, where Salesforce CEO Marc Benioff will open with Patrick and Daniel Newman. As AI moves from experimentation toward production, the industry needs operating evidence—not just demonstrations. 🤖 🔹 Monetization: Six earnings reports put AI economics under the microscope. NVIDIA will provide another read on demand, supply, financing and margins, while Marvell, Salesforce, HP, Everpure and Synopsys offer signals across custom silicon, enterprise AI, AI PCs, storage and chip design. 📈 Meanwhile, Bloomberg reports AI server prices are rising more than 15% amid higher memory costs—putting even more pressure on the economics of the AI buildout. And catch Patrick on Yahoo Finance Tuesday and CNBC Wednesday discussing NVIDIA. 📺 Watch The Week Ahead in Technology for August 24, 2026 for the full breakdown. 👇🏻

What’s ahead in tech this week? 👀 The AI bill is climbing—and the story is running on three clocks: chips coming, deployments landing, and earnings paying for both. ⏱️ In this week’s The Week Ahead in Technology, CEO & Chief Analyst, Patrick Moorhead, breaks down what he’s watching across the industry: 🔹 Chips: Hot Chips at Stanford brings one of the densest silicon disclosures of the year, with NVIDIA, AMD, Intel, Arm, Google, Microsoft, Meta, OpenAI and more presenting. The big takeaway? Custom silicon is increasingly sitting alongside the GPU—not replacing it. 🔹 Deployments: Patrick is co-hosting The Six Five Media Summit: AI Unleashed, where Salesforce CEO Marc Benioff will open with Patrick and Daniel Newman. As AI moves from experimentation toward production, the industry needs operating evidence—not just demonstrations. 🤖 🔹 Monetization: Six earnings reports put AI economics under the microscope. NVIDIA will provide another read on demand, supply, financing and margins, while Marvell, Salesforce, HP, Everpure and Synopsys offer signals across custom silicon, enterprise AI, AI PCs, storage and chip design. 📈 Meanwhile, Bloomberg reports AI server prices are rising more than 15% amid higher memory costs—putting even more pressure on the economics of the AI buildout. And catch Patrick on Yahoo Finance Tuesday and CNBC Wednesday discussing NVIDIA. 📺 Watch The Week Ahead in Technology for August 24, 2026 for the full breakdown. 👇🏻

What were our analysts watching across the tech landscape last week? 👀💡 From 🤖 AI agents and infrastructure to 🏭 silicon, 💾 storage, 📱 on-device AI, 🔐 governance, XR, edge computing, and more, the Moor Insights & Strategy team had plenty to weigh in on. In last week’s MI&S Weekly Analyst Insights, CEO & Chief Analyst, Patrick Moorhead, and VP & Principal Analysts Paul Smith-Goodson, Jason Andersen, Bill Curtis, Matthew Kimball, Michael Leone, Anshel Sag, and Melody Brue share their latest perspectives on the developments shaping the industry. A few topics from last week’s insights: 🔹 The potential risks of autonomous AI agents operating in swarms 🔹 The evolving agent governance layer 🔹 Oracle’s approach to governing AI-generated applications 🔹 AMD’s Taalas acquisition and the economics of AI inference 🔹 Intel’s capital raise and foundry ambitions 🔹 Developments in AI storage and edge AI inference 🔹 Google’s Pixel lineup and continued push toward on-device AI 🔹 The latest across XR, wearables, semiconductors, enterprise data, and more Plus, catch up on our analysts’ latest research, media appearances, podcasts, testing, and upcoming events. 🚀 Catch up on last week’s analyst insights here 👇🏻 https://lnkd.in/gRC8UBvr

The AI accelerator market is getting more diverse. But hardware choice only matters if developers can actually take advantage of it. 🧩⚡ In his latest Forbes analysis, CEO & Chief Analyst Patrick Moorhead looks at Modular, a Qualcomm company’s announcements from ModCon 2026 and the company’s push toward greater openness and accelerator portability. As AI infrastructure expands beyond any single accelerator architecture, software portability becomes an increasingly important part of the equation. Developers and enterprises want the ability to take advantage of the hardware that best fits a workload—without having to rebuild their software strategy every time the underlying compute changes. 🖥️🔄 That makes openness at the software layer potentially just as important as competition at the silicon layer. Patrick breaks down Modular’s approach, what he saw at ModCon 2026, and why accelerator portability could become increasingly important as heterogeneous AI infrastructure grows. Read his full Forbes analysis 👉🏻 https://lnkd.in/gBV8Jx3g

Industrial edge systems are getting smarter—and the humble MCU has a bigger job to do. 🏭⚡ In his latest Field Note, VP & Principal Analyst Bill Curtis examines NXP Semiconductors’s new MCX A5 MCU and its focus on bringing industrial Ethernet connectivity into the MCX portfolio. For industrial developers, connectivity, real-time control, security, efficiency, and increasingly intelligent processing all have to coexist within tightly constrained systems. The addition of industrial Ethernet reflects the growing demands being placed on MCUs as factories and other edge environments become more connected and software-defined. 🔗⚙️ Bill looks at where the MCX A5 fits within NXP’s broader MCU strategy and why its capabilities matter for the next generation of connected industrial systems. Read his analysis 👉🏻 https://lnkd.in/g_xxj_6Y

What happens when an already compelling AI PC gets another generation of refinement? 💻⚡ VP & Principal Analyst Anshel Sag takes a closer look at the updated Lenovo Yoga Slim 7x Gen 11—and finds a machine that improves on its predecessor in almost every way. The latest generation continues Lenovo’s push in the premium thin-and-light AI PC category, where performance is only part of the equation. Design, portability, battery life, display quality, connectivity, and the overall user experience all matter when determining whether an AI PC actually makes sense as an everyday machine. 🔋✨ In his latest Research Note, Anshel breaks down the improvements, the experience of using the new Yoga Slim 7x, and where there may still be room to improve. Read the full review 👉🏻 https://lnkd.in/gveezmP4