Why AI compute comes in rectangles: Modal CEO Erik Bernhardsson
In a Latent Space interview, Modal CEO Erik Bernhardsson discusses elastic GPUs, the growth of sandboxes and the economics of cloud infrastructure for AI agents.
Related stories
Liquid AI launches RIPL, a real-time inference engine for sub-50 ms LFMs
Liquid AI has released RIPL, an inference engine built for real-time use of its LFM models. On a single B200, LFM2.5-1.2B running on RIPL showed 14.4 to 28.4 times lower p95 latency than Qwen3.5-0.8B on vLLM, according to the company.
Google's Agent Compute Engine gives every Gemini agent its own hardware-isolated micro virtual machine
Google is introducing Agent Compute Engine, a runtime built specifically for AI agents that isolates each one in its own micro virtual machine. It is also broadening governance tools so companies can manage agents across cloud, hybrid and on-premises setups.
After Finland, France's Bull wants to build another European country's first AI gigafactory
Bull and Czech operator ČRA have signed a memorandum of understanding for an AI Gigafactory in the Czech Republic. The project, a candidate for the EU program, aims to provide sovereign AI to Central and Eastern Europe.
Meta stock sinks 6% as AI spend pushes free cash flow negative
Meta shares fell 6% after AI spending turned its free cash flow negative. The price of its five-year credit default swaps hit record highs, a sign that investors are growing worried about its debt.
Self-hosted LLM vs. cloud APIs: how to route between them
An n8n guide explains that running a model yourself is not always cheaper than calling an API. It covers when on-premises wins, what local hardware can handle and how to route requests between local and cloud models.