Trade Ideas
Meta signed a deal worth "double-digit billions" with AMD, which includes share warrants based on operational milestones. This is a massive validation stamp from a hyperscaler. The warrants align incentives, effectively turning a customer into a partner who benefits if AMD's stock rises (similar to the OpenAI deal). It proves AMD is a viable alternative to Nvidia for massive-scale inference workloads. The deal locks Meta and AMD together, providing AMD with guaranteed revenue visibility and a "strong affirmation" of their technology in a market dominated by Nvidia. AMD is still "playing catch-up" to Nvidia; the warrants only pay off if AMD hits aggressive targets; Nvidia remains the dominant incumbent with $50B/quarter revenue vs AMD's $10B.
Despite the AMD deal, Nvidia still has a separate deal with Meta, and its data center business is $50B/quarter (5x AMD's total revenue). The speaker notes "all boats are floating" and there is "no anxiety" yet because demand is so high that Meta buys everything available. However, the emergence of a multi-billion dollar competitor (AMD) and internal chips signals the start of margin pressure and market share erosion. Nvidia remains dominant, but the monopoly is cracking. The trade depends entirely on execution: if they miss targets, the narrative shifts to "competition is hurting them." Hyperscalers successfully diversifying away from Nvidia faster than anticipated; margin compression.
Mark Zuckerberg is "front-loading" purchases with a $135 billion Capex budget, buying from Nvidia, AMD, and building internal chips. Meta is aggressively securing the single most scarce resource in tech (compute). By diversifying suppliers (AMD/NVDA) and building in-house silicon, they reduce vendor lock-in risk and lower long-term costs. The "insatiable demand" signals they are building an insurmountable infrastructure moat. Meta is positioning itself as the best-capitalized player in the race to "super intelligence," securing hardware before shortages can hinder growth. Massive Capex spend could compress margins if AI monetization (revenue) lags behind the infrastructure build-out.
Riley
Tech Reporter/Analyst
Meta is targeting data center projects as large as 5 gigawatts but notes they need "energy companies to be able to deliver" and regulatory approvals. You cannot run $135 billion worth of chips without massive power. The transcript explicitly highlights the "merging of the energy and the compute." As chip supply loosens (via AMD/NVDA competition), the hard constraint shifts to the power grid and generation capacity. Utilities and Independent Power Producers (IPPs) hold the keys to the kingdom. Tech giants must pay premiums or fund infrastructure to get these gigawatts online. Regulatory delays; grid interconnection queues taking years; local pushback against data center energy consumption.
This Bloomberg Markets video, published February 24, 2026,
features Ian King, Riley
discussing AMD, NVDA, META, XLU.
4 trade ideas extracted by AI with direction and confidence scoring.
Speakers:
Ian King,
Riley
· Tickers:
AMD,
NVDA,
META,
XLU