Consensus on this seems to be ZDR which tracks to my own experience as well. If that really turns out to be the case, it's an incredibly strong and clear message to the labs on the impact of data retention policies.

Today in diminishing model returns

I was thrilled to join @SquawkCNBC for the first time this morning! For those monitoring: yes, we do very briefly touch on foregoing love in pursuit of AI.

They set a firewall on fire. Real flamethrower, real explosions. That's how @OPSWAT decided to explain why one firewall isn't enough for critical infrastructure. Free on YouTube: piped.video/LcYjfdefOZI #OPSWATPartner #ad

Alternative cloud provider IREN (Nasdaq: IREN) has officially closed its purchase of software maker @MirantisIT. According to SEC filings, IREN completed the all-stock transaction for over $500 million. Read up on our analysis (subscription required): futuriom.com/articles/news/i…

Minion by @nss_labs is a managed testing platform fthat evaluates whether deployed security controls effectively detect, block, allow, or mishandle defined threats and traffic. #cybersecurity Check out our latest #Techprimer: futuriom.com/articles/news/w…

#AI semi giant @AMD has joined up with @Supermicro and @spectrocloudinc to offer a system that enables developers using coding assistants for AI workloads to control the model assets they use, trimming token costs and realizing greater efficiency. futuriom.com/articles/news/a…

@futuriom Cisco Live coverage cuts straight to what makes FlowAI different: it's not another AIOps layer. FlowAI is built to take probabilistic AI + put real controls around the outcomes. Agentic Reasoning, Deterministic Execution, Connectivity. 1 engine. bit.ly/49NrCSN?utm_source=X&…

Not a good look for the #AI #Finance #Circularity

Iridium is putting Short Burst Data inside SKYWAVE (@ORBCOMM) terminals, sitting next to cellular and Wi-Fi under one management layer. Connectivity choice is moving off the modem and into software. That shifts who fleet operators buy from. iotbusinessnews.com/2026/07/…

The #AI finance circularity is having its moment in the market. @nvidia plans to invest hundreds of billions in speculative AI infrastructure, fueling fears that it may not pay off. #finance #bubbles futuriom.com/articles/news/n…

> @jpmorgan Chase's successful adoption of AI across its businesses highlights the divide between AI do-ers and enterprises running perpetual pilot projects. #AI #EnterpriseAI futuriom.com/articles/news/e…

Raynovich of @futuriom captured a shift underway in the neocloud market in his @RaiseSummit recap: as neoclouds evolve into mini hyperscalers, the differentiator is no longer GPU availability or price. It's operational excellence — data sovereignty, fault tolerance, and the automation required to deliver them at scale. Here at Netris, we are seeing the same shift. Neoclouds are building out GPU capacity at a massive rate, but most don't have the software depth the large hyperscalers spent years developing. Network orchestration and automation are where that gap gets closed. The article covers where companies like Netris fit: automating the network build-out of GPU clusters, abstracting the fabrics beneath them, and enforcing hard multi-tenancy in hardware. Read Scott's full analysis: futuriom.com/articles/news/n… #RAISESummit #AIInfrastructure #Neocloud

Hedge fund investor Michael Burry (@michaeljburry), who correctly called the 2027-2008 housing finance crisis and rose to fame after being featured in the book and movie The Big Short, is once again critiquing the circular financing deals backing the AI boom, specifically those featuring @nvidia and @OpenAI . $NVDA futuriom.com/articles/news/i…

The most expensive idle asset in the world? A GPU waiting on the network. When networking bottlenecks slow down your AI clusters, your ROI plummets. DriveNets just secured $410M in Series D funding to scale the open, multi-vendor Ethernet fabrics that large-scale AI infrastructure demands. drivenets.com/news-and-event… DriveNets’ high-performance AI Fabric eliminates networking bottlenecks by performing end-to-end networking optimizations across the entire AI stack. ✔️ Maximize GPU utilization ✔️ Reduce cost-per-token ✔️ Enable rapid deployment and efficient end-to-end scaling

Data management firm @qumulo Streamlines Hybrid Cloud Data Management for AI. #data #AIdata (Cloud Tracker Pro - subscription required) futuriom.com/articles/news/q…

With the launch of FlowAI, @Itential builds a portfolio on the cutting edge of agentic operations. #AIOps #AI #automation #networking futuriom.com/articles/news/i…

Nvidia Is Expanding Infra Partnerships. Will A Big Deal Happen? via @forbes forbes.com/sites/rscottrayno…

Is your network the hidden ceiling for your AI performance?​ While Mixture-of-Experts (MoE) architectures have drastically reduced compute costs, they have exposed a critical networking bottleneck that GPU investment alone cannot fix. ​ Unlike the predictable, choreographed communication of dense models, MoE creates "improvisational" and unpredictable traffic patterns that often lead to significant GPU underutilization. ​ 👉 drivenets.com/resources/whit… Explore how industry leaders like DeepSeek AI and Meta are already reporting that communication latency can account for up to 50% of training time or 30% of serving latency.  ​ The focus must now shift from raw processing power to optimizing the network fabric that connects it all.

How @cerebras became the largest chip IPO in history. $CBRS #AI #Chips futuriom.com/articles/news/c…