Table of Contents
Meta Platforms has entered into a massive infrastructure agreement with AMD to acquire six gigawatts of data center capacity and specialized AI hardware, a deal valued in the double-digit billions. The partnership, revealed amidst a broader "AI scare trade" that has sparked volatility across software valuations, represents Meta's most aggressive effort to date to diversify its compute supply chain away from Nvidia. This strategic pivot comes as social media giants scramble to secure the massive energy and silicon requirements necessary to reach the "super-intelligent" AI capabilities envisioned by CEO Mark Zuckerberg.
Key Points
- Meta has committed to a multi-year deal for six gigawatts of AMD chips and data center supplies, including a warrant structure that could make Meta a significant AMD shareholder upon meeting performance milestones.
- Anthropic’s unveiling of Claude Code triggered a sharp sell-off in legacy tech stocks, causing IBM to experience its steepest one-day share decline in 25 years due to fears of AI-led COBOL modernization.
- The Warner Bros. Discovery board is currently reviewing a sweetened, all-cash takeover proposal from Paramount as the battle for Hollywood consolidation intensifies.
- AI chip challenger Etched (referred to as NAI in recent funding rounds) secured $500 million to develop "blank slate" hardware specifically optimized for large language model (LLM) throughput.
Meta and AMD Challenge the Nvidia Hegemony
The deal between Meta and AMD is not merely a purchase agreement but a deep technical and financial integration. The contract focuses on AMD’s latest accelerators, specifically the MI350 series, and involves "co-engineering" future server designs to handle Meta's specific workloads. By including warrants in the deal, AMD has incentivized Meta to transition a larger share of its workloads—including its massive video and social feeds—to AMD silicon.
"This is a real strong affirmation and it also includes a share of warrants as well... This basically ties us closer to Meta. They get these shares if we are in a very good position. Her [Lisa Su's] point was that this kind of locks us together and tightens our relationship." — Ian King, Bloomberg News.
Industry analysts suggest this move targets the "inference" phase of AI, where models are deployed for user interaction. While Nvidia remains the dominant leader in AI training, Meta’s internal pipeline for custom silicon and its massive annual spend—projected at $135 billion this year—indicates a shift toward a multi-vendor environment to mitigate supply chain bottlenecks and escalating costs.
The "Scare Trade" and Legacy Software Disruption
The market's reaction to Anthropic's latest agentic tools illustrates the increasingly reactive nature of tech investing. When Anthropic announced that its Claude Code could modernize COBOL—a coding language dating back to 1959 that still underpins roughly 200 billion lines of global banking code—IBM shares plummeted. The sell-off highlights a growing fear that AI agents will not just assist, but eventually render legacy software platforms and the services that maintain them obsolete.
Market Asymmetry and Agentic AI
The volatility has extended beyond IBM to other application software vendors like Intuit and ServiceNow. Analysts describe a "shoot first, ask questions later" mentality among investors struggling to discern which companies will be cannibalized by AI and which will be enhanced by it. Hillary Fridge, a senior research analyst for I.T. services, suggests that the market is currently pricing in a "zero-sum game" where any gain for an AI agent is a direct loss for an incumbent vendor.
"It’s a scary time to be an application software vendor. The idea that software vendors lose the kind of pricing leverage they have enjoyed for so long because of that potential disruption is really something markets are struggling to grapple with right now." — Brody Ford, Bloomberg News.
Consolidation and New Hardware Frontiers
Beyond the chip wars, the media landscape remains in flux as Warner Bros. Discovery reviews an improved bid from Paramount. The deal, which has seen its offer price raised from an initial $30 per share, could reshape Hollywood's hierarchy as legacy players attempt to match the balance sheet strength of Netflix. At the same time, specialized AI startups are gaining ground in the private markets. Etched, a chip startup founded by Google alumni, recently raised $500 million to build hardware that prioritizes "computational density" and throughput specifically for Transformer-based models.
Reiner, CEO of the chip firm, noted that the goal is to create the "highest throughput per square millimeter of silicon" by breaking compatibility with legacy programming models that currently constrain established players like Nvidia. Similarly, AI-accounting firm Basis raised $100 million at a valuation exceeding $1 billion, signaling that "domain-specific" AI is becoming the new gold standard for venture capital.
Attention now turns to Nvidia’s upcoming earnings report, which will serve as a bellwether for whether the massive capital expenditures from companies like Meta are translating into sustainable industry growth. Furthermore, President Trump’s State of the Union address is expected to address the rising energy costs associated with AI data centers, potentially signaling new negotiated commitments for Big Tech companies to shoulder more of the national electricity burden.
