VCs still pour money into AI tooling
Reports show venture capital continuing to flow into AI startups and hardware‑adjacent tooling, reinforcing urgency for pilots and capacity commitments across compute and cooling stacks (visualcapitalist.com) (fortune.com). The funding environment is keeping demand pressure on GPU and infra vendors.
AI deals made up 52% of global VC deal value in Q4 2025 (visualcapitalist.com). Venture investors poured about $192.7 billion into AI startups across recent reporting, according to Bloomberg-cited industry tallies. (forbes.com) Big infrastructure cheques are headline-makers: Databricks closed a $5 billion raise in February 2026 at a $134 billion valuation. (cnbc.com) Lambda (the GPU/cloud provider) raised a reported $1.5 billion in November 2025 to scale AI data‑center capacity. (techcrunch.com) NVIDIA moved capital into the supply chain — a $2 billion strategic investment in CoreWeave was announced in January 2026 to accelerate buildout of AI capacity. (bloomberg.com) CoreWeave and partners are targeting more than 5 gigawatts of AI factories by 2030 under expanded deals with NVIDIA. (investors.coreweave.com) Physical bottlenecks are driving procurement timelines: industry analysis estimates CoWoS advanced‑packaging and HBM memory were the primary constraints on AI‑accelerator output in 2025. (epoch.ai) TSMC has been accelerating AP7/AP8 packaging ramps but packaging capacity remains tight through 2026 in multiple supply‑chain reports. (trendforce.com) Cooling and power are moving from edge concern to central line item — NVIDIA confirmed next‑gen DGX systems will require liquid cooling, and market research projects liquid‑cooling penetration in AI data centers to exceed 30% as high‑power GPUs scale. (datacenterdynamics.com) Schneider Electric and other vendors are publishing liquid‑cooling deployment guides to meet those density requirements. (blog.se.com) Hyperscaler silicon is a live competitive force: AWS reports Trainium/Inferentia deployments at scale and has positioned its custom chips as a multibillion‑dollar business, while Google has expanded TPU capacity and signed large TPU deals with AI firms. (aws.amazon.com) These cloud‑native ASIC moves are shaping startups’ cloud vs. on‑prem buying patterns and total‑cost comparisons. (tomshardware.com)