Joe Flynn

"What has become increasingly clear over the last 24 months is that ... AI has become capable of performing what looks more like the work that humans do — versus the software tools they use. So if AI is moving from tool to worker and starting to do jobs itself, this is beyond the current "application layer." These are systems that can plan, act, think, reason, and actually get real work done. And when companies sell these types of systems, they are selling into labor budgets of customers — not just software budgets. This is what we think is the new frontier for the next few years." Thank you Kristian Marquez, CFA for capturing our approach so well!

23

If you missed enterprise cyber and AI storage leader Infinidat at Gartner IOCS last week, join us TOMORROW (Dec 18) as we discuss the critical trends in enterprise storage at the 7th Annual Insight Jam Conference in our keynote address: "Key Enterprise Storage Topics For Machine Age Market Architecture". Tune in at 8am PT/11am/4pm GMT/5pm CET to explore how 2026 will unfold for enterprise storage: • Cyber storage resilience and its critical importance to an overall corporate cyber security strategy • AI - both enterprise storage for AI applications and workloads and also the use of AI inside of storage systems for ease of management and to cut costs • The impact of enterprise storage on power budgets and power efficiency • How enterprise storage can dramatically reduce IT CAPEX, IT OPEX, and IT resource needs https://bit.ly/48T0osy Oren Yesharim David Beltrán Vera David Humes II David Sawyer David Muench Tim Dales Bill Basinas Fran Bruno David Lubrano Benjamin Lew Tom Manganaro Tom Dionne Daniel Potts Donato Ceccomancini Marco Meraviglia ☁ Andrea Sappia Andrea Fraietta Marcus Benham Juliet McMullen Anthony DeLisio Karine Haase Lynsy (Calvario) Marshall Rebecca Kraushaar Revital Aharon-Frimerman Kfir Kayakov Shira Ginsburg Solutions Review Tim King Maleeha Haider Jonathan Paula 📸Craig Hibbert Bob Stone Pete Byrne Paul S. Gaccione Paul Stewart George Deveau James Meyers Gene Kussmann JT Lewis Lee Bushnell

5

Come see us at Enterprise Connect in booth #928 to discuss AI. This announcement is an important milestone towards Nectar's ability to monitor and alert on your AI communications infrastructure. More press releases coming!

4

Infrastructure Built for Your Workload, Not Your Vendor's Margin Most vendors optimize for margin, costing you tens of thousands of dollars annually in wasted capacity. The HPC infrastructure market prioritizes vendor profit over customer outcomes. Resellers stockpile identical configurations, focusing on volume and price, while your actual computing problem remains secondary. The Hidden Costs of Generic Solutions Standard server setups run at 70% capacity. The real waste isn't hardware; it's in missed opportunities: ✓ Slower model training  ✓ Inefficient inference  ✓ Compromised power budgets  ✓ Memory bandwidth bottlenecks These issues arise when infrastructure doesn't align with workload. Why This Matters Now 79% of decision-makers now prioritize solution efficiency and total value over capital cost. Custom infrastructure that improves efficiency by 15% recovers its investment in under 18 months. On-premises solutions outperform cloud by 40-60%, if tailored correctly. What Workload-Driven Architecture Looks Like A consultative approach delivers: ✓ Memory Bandwidth Alignment – Optimized data flow with no waste.  ✓ Optimized Cooling – Tailored solutions for hot and cool workloads.  ✓ Efficient Network Topology – Multi-GPU training without bottlenecks, saving 20-30% in network costs.  ✓ Scalable Design – Built for growth without starting over. The Vendor Relationship Truth Many vendor relationships are transactional, leading to bottlenecks and lock-in. A consultative vendor challenges you to think bigger, sometimes recommending less hardware to avoid overprovisioning. This honesty is rare and transforms partnerships. Why We Built Malogica Systems Differently We recognized the pattern: vendors prioritizing margins over workload. We asked one question: What if we analyzed YOUR infrastructure needs BEFORE recommending hardware? Everything we do follows from that: ✓ We understand your AI models and data flow  ✓ We analyze before we recommend (not recommend then hope)  ✓ Sometimes fewer GPUs (not more)  ✓ Sometimes different cooling  ✓ Sometimes hybrid approach you hadn't considered Infrastructure that matches your workload. Not our margin. The Real Question Who truly understands your workload enough to tell you what you actually need? If your current vendor can't answer that, you're likely overpaying for misaligned infrastructure. We analyze first. Design second. Recommend third. If you're evaluating infrastructure vendors, especially in the planning phase, let's discuss your actual needs.  #AIInfrastructure #EnterpriseCompute #TechStrategy

10

3 Comments

The AI revolution demands a seamless, unified ecosystem to unleash its full potential. The fragmented landscape of software and hardware has long been a bottleneck, until now. I’m thrilled to celebrate Chris Lattner, Tim Davis, and the exceptional Modular team on their latest funding round, a testament to their audacious mission to redefine AI development. Chris and Tim possess a rare combination of deep technical expertise and an unwavering commitment to solving one of the most significant challenges in technology. At GV (Google Ventures), we invest in founders as much as their vision, and Chris is a true force of nature. As I shared with WIRED’s Lauren Goode, he’s fiercely opinionated, relentlessly driven, and often brilliantly right - a combination with industry-shaping potential. The Modular team is addressing the complexity of AI development head-on, building what Chris aptly calls the “Android for generative AI hardware.” Much more pithy than what I've referred to as "Kubernetes for chips". They are creating the unified, high-performance AI engine the industry has been waiting for, a platform that allows increasingly-heterogeneous hardware and software to operate as one. We at GV have proudly backed Modular since leading their Series A, and are thrilled to join their Series C. This new capital will fuel their mission to scale the platform, grow their world-class team, and continue to deliver the infrastructure shaping the future of AI development globally. Here’s to Modular! They’re not just building a company, they're laying the foundations for a platform that will define how AI is built and deployed for decades to come. 🚀

150

12 Comments

Comcast is at a critical juncture as it prepares to spin off cable networks into Versant, contends with a streaming platform that lacks scale domestically and is not available outside the US, and leans on Parks for growth amid broadband pressure from fiber and fixed wireless. Leadership transition adds uncertainty, with Mike Cavanagh positioned as potential sole CEO. In our view, acquiring Warner Bros.' studio and streaming assets offers strategic and potentially financial upside-strengthening Peacock's economics, enhancing the IP flywheel, and creating a credible path to multi-year FCF/share growth post-integration. While regulatory hurdles, tech-stack integration, and brand integration pose questions/risks, we view the risks of losing out on WBD to be far greater to CMCSA. In short, Comcast risks being left behind as PSKY or NFLX scale DTC, limiting Peacock's reach and weakening NBC's ability to compete in the global media market over time.

3

Exciting times ahead in IT distribution! GTDC's CEO Frank Vitagliano highlights how agentic AI and the emergence of frontier partners are driving transformative changes. Discover how these innovations will reshape our industry: https://okt.to/qXd6Tt

3

The irony of the Technology Channel is that we are painfully behind on using technology to fix our own sales and discovery processes. We talk a big game about AI and business outcomes, yet we are constrained by archaic discovery methods. We are still relying on subjective interviews and screaming 1995 era forms to build generic, 135 page reports that nobody actually reads. Right now, a staggering 90% of AI implementations fail because organizations try to boil the ocean without a blueprint of their granular workflows. There is a massive visibility gap between how leadership believes work happens and how it actually happens. The missing piece? Context. - You cannot deploy an AI Agent to do a job if you don't actually understand the job. - Without capturing the undocumented tribal knowledge hidden in your employees' heads, AI just exposes how much of your business was never written down. - If an AI lacks the context (the why behind the raw data), it will just produce generic hallucinations. It is time we shift our industry from subjective opinions to objective data observation. We need to stop guessing what the client needs based on manual interviews, and start capturing the ground truth to fix real business bottlenecks.

13

1 Comment

Demand response participation is increasing as a result of increased DER adoption due to resiliency needs and long interconnection queues. The lost revenue due to not being powered is a bigger driver than capturing demand response revenue. I don't think there's an argument that demand response works. I do think there's an argument to be made that it's difficult to capture value from demand response alone. For that reason, I've argued that demand response products will have to be features of larger platforms. https://lnkd.in/gueF2QnK

12

2 Comments