On the Ground at UBS’ Private AI, Software and Internet Conference | Menlo Park

The software and AI sectors are undergoing a significant reallocation of capital as enterprises prioritize AI infrastructure and security over traditional application software. While major cloud providers face capacity constraints, startups are increasingly turning to specialized "neocloud" providers to secure necessary NVIDIA GPU access. Financial institutions, such as UBS, are navigating this shift by implementing "model gardens" that route tasks to the most efficient language models, emphasizing safety and trust over raw frontier model performance. Meanwhile, the robotics industry faces a challenge regarding real-world data collection, shifting focus toward autonomous warehouse solutions that provide immediate operational efficiency. Despite the hype surrounding frontier models, the market is pricing in a long-term threat to incumbent software firms as enterprises explore custom-built AI alternatives, though widespread replacement remains a future prospect rather than a current reality.