An interesting read below:
AI may have begun as a race to build ever-larger models, but the battle is increasingly shifting to the infrastructure underneath them. Last week brought two notable reminders: Qualcomm’s nearly $4 billion acquisition of AI software startup Modular and reports that chipmaker SambaNova is finalizing an $800 million funding round at a $10 billion valuation. In an interview, GV investor Dave Munichiello speaks with Crunchbase News on a range of topics including why the software layer connecting increasingly diverse AI hardware is becoming strategically valuable, what Qualcomm’s Modular acquisition signals for other startups, and why he still sees a path to blockbuster IPOs in AI infrastructure. Plus, it’s not just the Modular deal: Crunchbase data shows a broad resurgence in billion-dollar startup exits, we revisit last week’s biggest U.S. funding rounds, and an investor makes the case that venture capital needs better data.
A GV investor on its 10x Modular return and AI’s next shift
As demand for AI inference explodes, GV managing partner Dave Munichiello says the next wave of infrastructure startups will be defined by efficiency: getting more value out of scarce, expensive compute. In an interview with Crunchbase News, Munichiello discusses why AI workloads are moving toward “disaggregated inference,” how open-source models could change who buys and runs AI infrastructure, and why he believes hardware-heavy startups can still grow into large, independent public companies. He also shares what the firm learned from its early investment in Modular, for which it is poised to earn a 10x return on dollars invested with the startup’s acquisition by Qualcomm announced last week.