Nscale closed a $3.36B pre-IPO round led by Third Point, with $1B from Nvidia still to come
THE SO WHAT
Pre-IPO convertibles of this size mean AI infra is being financed like long-duration industrial buildout, not typical SaaS growth. If you depend on third-party compute, assume more capacity will be locked up in quasi-captive structures and start mapping your exposure to these capital alliances.
READ THE SOURCE
MORE FROM THE WIRE
Startups & VentureAccel Sees AI Opportunity Shifting to Applications
When late-stage investors like Accel say the AI upside is moving from infra to applications, they’re telegraphing where the next valuation premiums will accrue. If you’re still pitching “horizontal AI platform,” expect tougher questions — workflow depth, distribution, and clear payback are about to matter more than model specs.
Startups & VentureThe Week’s 10 Biggest Funding Rounds: Cybersecurity, AI And Health Take The Lead
Two $400M cybersecurity rounds alongside big checks for foundational AI, drug discovery, and neurotech show capital clustering around “AI plus regulated surface area.” If you’re raising, sharpen your story on how AI touches either security or health-grade outcomes — that’s where late-stage money is leaning.
Startups & VentureSources: SK Hynix's US-based NAND and SSD subsidiary Solidigm is exploring an IPO as early as 2027 that could raise $15B and value the unit at up to $150B
A potential $15B raise at up to $150B valuation for Solidigm underlines how storage is being repriced as AI-era infrastructure, not commodity hardware. If your workloads are storage-heavy—RAG, logs, video—assume both cost and strategic importance of NAND/SSD supply will climb and negotiate longer-term terms where you can.
Startups & VentureSK Hynix’s Solidigm Is Said to Weigh US IPO as Soon as 2027
Solidigm eyeing a US IPO as soon as 2027 reinforces that NAND and SSD capacity are strategic assets in the AI buildout. Infra buyers should treat storage vendor concentration and long-term availability as seriously as GPU supply when planning data-heavy architectures.