Colin VanderSmith

Most of you think of OpenAI as a pure software play. But what most of you don't know: they launched a consulting arm this summer with $10M minimum engagements. This embeds engineers directly into client organizations. So even the company that created ChatGPT can't just sell technology. They need humans to make it actually work. Think about that!! Guys, the most successful AI company in the world just admitted that software without implementation is worthless. This creates MASSIVE opportunity for builders like us. While OpenAI burns billions trying to figure out how to deliver value, businesses are desperate for working solutions right now. They don't want access to APIs. They want apps that solve problems. We've been doing this the whole time: taking technology and turning it into real business value. Speed wins. Results win. 🚀

34

18 Comments

Yesterday, the IBM acquisition of Confluent officially closed...and if the Red Hat acquisition taught us anything, it’s that $11B valuations usually lead to "Cloud Pak" bundles and gated source code. As Jeff Mery breaks down in his latest deep dive, the "IBM Cycle" often trades community-driven innovation for high-margin proprietary features and aggressive audits. If you’re tired of the "Streaming Tax" and want to see how 100% upstream, diskless Kafka can slash your infra costs by 97% before your roadmap gets "IBM-ified," this is the one article you need to read today. Read the full analysis and see the breakdown here: https://lnkd.in/dZ9bBt5V

41

Sovereignty and AI: Innovation without the trade-offs. For too long, organizations have been forced to choose between two speeds: moving fast with AI-powered modernization or slowing down to meet stringent sovereignty requirements. Through our expanded partnership with Capgemini, we’re proving that the trade-off is over. By combining Google Cloud’s Vertex AI and Gemini with Capgemini’s deep regulatory expertise, we’re helping global organizations: - Modernize with Confidence: Scale AI-driven innovation within sovereign boundaries. - Exercise Choice: Tailor sovereignty models across Google Cloud, Google Cloud Dedicated, and GDC air-gapped offerings. - Engineer Trust: Building protection into the foundation of modern cloud and AI platforms, not just bolting it on. Read this latest blog from Herschel Parikh at Capgemini on how we’re turning sovereign ambition into operational reality. 🛡️☁️

7

https://lnkd.in/gzebwkkW Description Swami Sivasubramanian, AWS | AWS re:Invent 2025 In this interview from AWS re:Invent 2025, Swami Sivasubramanian, head of Agentic AI for AWS, joins theCUBE’s John Furrier to discuss the seismic shift toward agentic AI. Sivasubramanian frames this evolution as one of the most significant transformations in tech history, positioning it not just as a productivity booster but as a fundamental reinvention of software development. The conversation highlights the launch of Frontier agents and NovaForge, noting how AWS is moving beyond basic code generation to address the full development lifecycle. Sivasubramanian points out that actual coding is only 20-30% of a developer's job, with the rest consumed by DevOps and SRE tasks. He details how virtual teammates can now handle these complex workflows, citing a powerful metric where a six-person team built a major inference system in just 76 days – a project that traditionally would have taken 30 to 40 people a full year. The dialogue explores the distinction between standard AI assistants and true "teammates" that possess long-term memory and understand unique team preferences, effectively moving beyond the "day one intern" model. Sivasubramanian elaborates on the success of Strands, an open-source framework that has rapidly accrued five million downloads, empowering developers to build agents without complex orchestration logic. Furrier and Sivasubramanian also examine the infrastructure implications of this shift, discussing how innovations in Trainium and vector storage are critical for handling the massive influx of telemetry and data associated with agentic workflows. The segment concludes with a look at modernization, where agents are turning years-long legacy migrations into projects measured in months, setting the stage for a future where every application is agentic by default.

2

We're excited to announce new capabilities in Google Cloud #VertexAI Training designed to simplify and accelerate large-scale model development. Key highlights from our latest update: 🔹 Flexible, Self-Healing Infrastructure: Leverage fully managed, resilient Slurm environments with automated failure detection and performance-optimized checkpointing. 🔹 Cost-Effective Scheduling: Utilize Dynamic Workload Scheduler (DWS) for fixed future reservations or flexible on-demand capacity. 🔹 Integrated Frameworks: Access optimized recipes for the full model lifecycle (including SFT and DPO) and seamless integration with NVIDIA NeMo. Whether you're fine-tuning standard models or training massive custom foundational models from scratch, these new features are built to get you to production faster. Read the full announcement to see how organizations like Salesforce and AI Singapore are already leveraging these tools to elevate their models. 👇 https://lnkd.in/gVVq3WCw

185

2 Comments