Private Capital Is Plugging State Spending Gap, Apollo's Zelter Says
THE SO WHAT
If private capital is stepping in where state balance sheets can’t support large-scale industrial capex, the control points for infrastructure — including compute, energy, and manufacturing — will tilt further toward asset managers. Operators planning big physical or AI builds should expect more private-credit style terms and covenants attached to their core infrastructure.
READ THE SOURCE
MORE FROM THE WIRE
Startups & VentureCustom AI chip design startup TYLsemi raised $43M in early-stage funding led by Matter Venture Partners to build modular chips thanks to packaging tech advances
Modular AI chiplets with advanced packaging are a bet that customization, not just scale, will matter in the next compute cycle. If you’re planning multi-year AI infra, start tracking chiplet ecosystems and packaging roadmaps alongside the usual GPU SKUs.
Startups & Venture‘Listing is a must’: China’s LimX raises $200m as the humanoid IPO race heats up
A $200M raise at a $2.21B valuation with IPO pressure shows humanoids are being financed like a capital race, not a slow-burn R&D project. If you’re in logistics or manufacturing, expect aggressive pilots and pricing from multiple humanoid vendors over the next 12–24 months.
Startups & VentureSo Far, 2026 Is A Solid Year For Cybersecurity Startup Funding
$4.4B in cyber and privacy funding in Q2 — even with a ~30% drop from Q1 and last year — shows security remains one of the few categories where venture is still writing large checks. If you’re an enterprise buyer, expect more vendor noise and overlapping claims; tighten your evaluation criteria around integration and measurable risk reduction.
Startups & VentureDeepSeek Mulls New Funding Weeks After $7 Billion Round, FT Says
DeepSeek exploring more capital just weeks after a $7B round shows how capital-intensive frontier model races have become — balance sheets are being built for multi-year burn, not single cycles. For smaller AI players, this is a cue to avoid head-on model wars and instead anchor on distribution, domain data, or workflow depth.