Rob Strechay at the next place

Everyone is talking about AI models. But after spending time at @VAST_Data's VAST Forward, one thing became very clear: Agentic AI isn’t a model problem. It’s a data platform architecture problem. As organizations move beyond copilots into AI systems that take action, the real bottleneck isn’t the LLM. It’s the platform underneath: • governed data access • high-speed data platform services • persistent AI memory • GPU-accelerated data services • distributed infrastructure that can manage fleets of AI systems At VAST Forward, the conversation wasn’t about bigger models. It was about how the data platform becomes the runtime for AI agents. That shift is huge. The companies that win in the next wave of enterprise AI won’t just deploy better models; they’ll build data platforms designed for agentic systems. I break down the four architectural shifts emerging from VAST Forward and what they mean for platform teams, CTOs, and data engineers. 👇 Short video (6 min) and Full article below linkedin.com/pulse/agentic-…

Context is Key to Agentic AI: Listen to Where Code Meets Capital on Spotify for Creators open.spotify.com/episode/4pTetQ…

Most organizations think Agentic AI is a model problem. It isn’t. It’s a context, governance, and data architecture problem. linkedin.com/pulse/context-… In my latest conversation with @jaswani from @starburstdata we dive into why semantic layers, metadata, and contextual intelligence are becoming the real control plane for enterprise AI. This ties directly into the momentum behind the Dell AI Data Platform with NVIDIA and where the market is heading next with Agentic AI and Sovereignty.

@JClarkeatDell @DellTech Dell Private Cloud is all about simplification of deployment and management Typical HCI doesn't get you this, because of stair step cost - storage scaling separately is key Multiple flavors with @openshift @nutanix ahv, and @VMware VCF 9.1 Dell Automation platform is key!

Top Ten changes in the past 12 months... DellTechnologyWorld Day 2 with @JClarkeatDell @DellTech Thread 👇 1/ AI as operator, not mere advisor 2/ model prices crater 3/ token consumption up 10x - seems low 4/ context windows cross millions of tokens 5/ train built models, now inference runs them 6/ GenAI spend tripled last year 7/ physical AI in production 8/ PC part of AI stack 9/ talent question flipped 10/ conversation changed

@JClarkeatDell @DellTech Protection for AI and the data is going ro be key. Air-gapped security for backups will be key, like PowerProtect One - hardware root of trust and zero trust access - now driving down management overhead and massive data reduction.

@JClarkeatDell @DellTech Also Powers ale makes an appearance - talking Nvidia certification. Next up is ObjectScale with RDMA over S3 - adding in S3 tables too. Lastly Exascale with all 4 software defined storage in a rack - this makes sense as I talked to @PariseauTT about - but optimization will be key

Data is the life blood of AI - Arthur Lewis discusses the amount of data on-premises - Dell AI Data Platform using APIs and SQL across iceberg to bring structured and unstructured data to the AI Talking about Lightening FS to help disaggregated inference...

Most efficient AI is using the token produced closest to the data! This makes sense. GB10 with @nvidia nemo claw can save significantly

Now @openclaw discussion - Jeff about how Agents are changing productivity. Security and O11y are super important and a major fear of organizations. 1/ read-only 2/ O11y so you know what it is doing Challenge is connecting data.

Imperatives for the AI enterprise

Observations 1/Token costs are falling fast. Token use is exploding! 2/Productivity for individuals using AI is exponential 3/Tokens are a line item in P&L

@sarbjeetjohal @Apple Did you see the new @awscloud Ultra Mac instances? Announcing general availability of Amazon EC2 M3 Ultra Mac instances - AWS share.google/0F2hmdS9UkjGyT…

First of all I think that's temp phenomenon... second is that I focus more on software side... having said that, software developers will benefit from such facility but I still think that it belongs in Cloud, a VM or a set of VMs and/or containers. Me leaving my laptop or desktop running at mercy of local wires and power to complete a a set of tasks is stone age computing, IMHO :) In Other words, this discipline will mature with time. In fact, on hardware side, I would love to see Apple's native silicon (server side) and their support for devs in their "Cloud", at some point!

.@Apple’s WWDC happens frequently m June 8-12 at Apple Park in Cupertino, California. My #WWDC participation is still up in the air, no pun intended 😊! I am looking forward to coverage from @BenBajarin!

@sarbjeetjohal @Apple What about hardware wise? Lots of Mac Minis being used with OpenClaw and more. But doesn't really have a "stack" and harness outside some more intense ones...

.@RealStrech I am glad you asekd :) These are 4 major things I am expecting from @Apple WWDC: 1. Overhaul of Siri to thinking chatbot like functionality 2. I expect AI integration across all native applications, web search, with cross-device functionality. 3. Native IDE Xcode getting better AI code generation and spec driven development. Support for long running jobs. 4. After OpenAI - Apple fallout, I expect Apple-Google partnership to deepen.

@QuinnyPig You are on 🔥❤️‍🔥 today, sir!

All the inference providers charging suspiciously similar per-token rates while insisting they're fiercely competitive? This is called a "Token Ring Network.”

@SavIsSavvy @nvidia @RedHat They, @MistralAI, were on the logo board but were talking more about the others they were adding.

@RealStrech @nvidia @RedHat Great thread Rob! I’m curious to hear what they say about Mistral

AI and Agents reinventing industries is a major theme for @MichaelDell @DellTech DellTechnologyWorld First on stage is Eli Lilly talking about technology scaling pharma Thread inbound 👇

11/ @PariseauTT's deep dive on @DellTech Day 1: Strechay said he expects agentic AIOps, which Dell also expanded for infrastructure management this week, to play a role in cross-vendor and cross-site data center orchestration. "I would assume in the future the automation will obfuscate different systems through an MCP server or API set that will enable AI to make the determination of where the workload should sit," he said. techtarget.com/searchdatacent…

@nvidia @RedHat MORE READING 👇 10/ For more on Token Economics, check out my article with data from @VentureBeat and a video with the CTO for AI of @LinkedIn venturebeat.com/infrastructure…

8/ would it be a DellTechnologyWorld without @nvidia Jenson Huang? Nope. Talking strategy of organizations doing Agentic AI at scale. Compression of innovation is his big topic theme.