Nvidia and SK Hynix signed a $500 billion memory supply agreement, locking in AI-driven chip demand for years. Samsung and Broadcom followed with a $200 billion MOU, part of a broader roughly $950 billion South Korea-US AI infrastructure package.1
The US government took an equity stake in GlobalFoundries, marking direct state involvement in domestic chip capacity. For investors, that shifts GlobalFoundries from a pure-play foundry bet to one partly backed by federal policy support.
Consolidation is also moving through mergers. Skyworks and Qorvo announced their combined leadership team, with Skyworks CEO Bob Bruggeworth citing a partnership that has "shaped our integration planning efforts from the very beginning."2 Qorvo CEO Phil Brace joins him in steering the merged RF chipmaker through its integration phase.3
Elsewhere in the supply chain, Kalray and Bull (Eviden) announced a partnership to develop high-speed networking for next-generation AI and HPC infrastructure. Kalray CEO Éric Baissus said the deal "confirms the relevance of our technological vision" as AI reshapes infrastructure needs across sectors.4
Packaging and test specialist Amkor Technology reported results reflecting the same capacity buildout, as AI chip volumes push demand through the back end of the supply chain.5
For traders, the pattern is clear: capital is concentrating around firms that can lock in AI memory and chip capacity years in advance. Nvidia and Samsung's mega-deals set pricing and supply terms that ripple through smaller suppliers, while M&A activity like Skyworks-Qorvo signals scale is becoming a competitive necessity, not an option. Agentic AI tools now embedded directly into chip design workflows with Cadence, Siemens, Synopsys and Silvaco add a second layer of consolidation pressure, tying design-software vendors closer to the same hyperscaler capital driving the memory and foundry deals. State intervention via the GlobalFoundries stake suggests policy risk and support are now inseparable from valuation in this sector.
