News Summary:
On April 29, 2026, Val Bercovici of Weka discussed "tokenmaxxing versus signalmaxxing" in AI, referencing recent Wall Street Journal and i-SCOOP articles that debate optimizing signal per token versus burning more tokens. Earlier, on April 22, Bercovici, Chief AI Officer at Weka, spoke with SDxCentral's Kat Sullivan about overcoming the "AI memory wall" and supporting scalable large language model (LLM) and agent swarm infrastructure without compromising performance. The same day, Weka participated in an ISC Executive Insights Dinner focused on "AI Without Compromise: Performance. Trust. Sovereignty," highlighting AI's role as national infrastructure and the challenges for emerging "AI factories" regarding control, cost efficiency, sovereignty, and scalability. This followed Weka's involvement in a "Vendor Showdown" at ISC 2026, which explored the future of data infrastructure for modern HPC and AI workloads under strict power, cost, footprint, and sovereignty constraints. Previously, on April 21, Bercovici explained Weka's role in providing "storage and memory software" for AI infrastructure to address challenges like the "IO Blender and Memory Wall."
Subscribe for full access to Weka's products in full detail 
