Meta, AMD Deal Fuels AI Spending Surge | Open Interest 2/24/2026
Speakers
Summary
- AI "Feedback Loop" Anxiety: A widely circulated research report suggests a negative feedback loop where companies fire workers to fund AI, which increases efficiency, leading to more layoffs and weaker consumption. This "AI anxiety" is causing volatility, specifically punishing legacy tech (IBM) while rewarding AI adopters.
- Chip Wars Intensify: The monopoly narrative is cracking. Meta signed a multi-billion dollar deal with AMD, validating them as a true alternative to Nvidia for inference. Meanwhile, SambaNova and Intel are pushing new architectures claiming 5x speed advantages.
- Physical over Digital: Goldman Sachs argues the "AI trade" is rotating. The "Mag-7" trade is crowded/flat, while capital is flowing into the "Physical" economy (Industrials, Energy, International) required to build the AI infrastructure (power, cooling, data centers).
- Energy Bottleneck: The CEO of Baker Hughes confirms that while oil supply is high, the real story is Natural Gas. AI data centers are driving massive demand for gas turbines for baseload power, with 2026-2027 projected as a major growth cycle.
- Macro Backdrop: The Fed (Goolsbee) is hesitant to cut rates further until inflation (stuck at 3%) breaks, despite a steady labor market. Simultaneously, Trump's tariffs create uncertainty for importers, pushing investors toward domestic industrials.
Trade Ideas
IBM shares dropped 13% (worst day in 25 years) after Anthropic released a tool that can modernize/rewrite COBOL code. IBM's "moat" is largely built on legacy mainframe systems running COBOL (200 billion lines of code in finance/banking). If AI can easily translate this to modern languages, IBM's sticky legacy revenue evaporates. SHORT. The market perceives AI as an existential threat to IBM's consulting and legacy maintenance business. Market overreaction; IBM is also deploying its own AI (WatsonX) which could offset losses.
Anthropic (makers of Claude) announced partnerships with these specific companies. Intuit is using it for tax AI; Spotify for efficiency; Novo Nordisk for research. The market is currently rewarding "AI Adopters" who integrate best-in-class models to reduce headcount/costs (the "efficiency" narrative). These partnerships signal these firms are successfully navigating the AI transition rather than being displaced by it. LONG. These stocks are bouncing on the "AI partnership" halo effect. If AI integration becomes a commodity, these companies lose pricing power (the "race to the bottom" on software pricing).
Goldman Sachs notes the S&P 500/Mag-7 are flat/down, while Industrials and International markets are up ~10%. We are seeing a cyclical rotation from "Digital" (pure software/tech) to "Physical" (the companies building the data centers, power grids, and hardware). Additionally, international markets offer better valuations and diversification against US tariff uncertainty. LONG. Focus on the "builders" of the AI economy rather than just the model makers. Global recession or escalation in trade wars (tariffs) hurting global industrial demand.
Meta has committed to buying billions of dollars in AMD chips over five years for their new 6-gigawatt AI factories, specifically for inference tasks. This is the "validation event" the market was waiting for. It proves AMD's MI300 series is a viable alternative to Nvidia for hyperscalers, breaking the single-vendor lock-in narrative. Meta gets cheaper compute; AMD gets massive volume. LONG. This re-rates AMD as a serious competitor in the inference market. Execution risks on AMD's software stack (ROCm) versus Nvidia's CUDA.
Baker Hughes CEO states that AI data centers are driving a surge in demand for gas turbines and power equipment. He sees 2026 as a trough before a massive 2027 recovery. Renewables (solar/wind) cannot provide the 24/7 baseload power required for gigawatt-scale AI factories. Natural Gas is the only scalable, immediate solution. This creates a structural bid for gas infrastructure and commodities. LONG. Energy infrastructure is the bottleneck for AI scaling. Regulatory pushback against fossil fuel expansion for data centers.
Rodrigo Liang
CEO, SambaNova
SambaNova (AI chip startup) is partnering with Intel to deploy its new SN50 chips, which claim to be 5x faster than rivals. Intel is desperate for a win in the AI foundry/partnership space. If SambaNova's technology succeeds, it pulls Intel back into the conversation as a manufacturing partner for next-gen AI silicon. WATCH. Speculative turnaround play if the partnership yields volume. Intel's execution history is poor; Nvidia's dominance is entrenched.
Up Next
This Bloomberg Markets video, published February 24, 2026, features Dani Burger, Caroline Hyde, Katherine Bordlemay, Ed Ludlow, Lorenzo Simonelli, Rodrigo Liang discussing IBM, CRM, INTU, SPOT, NVO, DOCU, XLI, EFA, AMD, META, UNG, XLE, INTC. 6 trade ideas extracted by AI with direction and confidence scoring.
Speakers: Dani Burger, Caroline Hyde, Katherine Bordlemay, Ed Ludlow, Lorenzo Simonelli, Rodrigo Liang · Tickers: IBM, CRM, INTU, SPOT, NVO, DOCU, XLI, EFA, AMD, META, UNG, XLE, INTC