Here are Friday's biggest analyst calls: Nvidia, SpaceX, Apple, Tesla, Meta, AMD, BP, Costco, Nike & more

Wall Street's Friday calls skewed positive, with new buy/overweight ratings and upgrades across Johnson Controls, Chemed, Fervo Energy, Stem, BP, AMD, Meta, Humana, Global Payments and others. Bank of America raised AMD's price target to $720 from $620, citing higher CPU pricing from agentic-AI demand, while Deutsche Bank lifted Meta's target to $820 from $750 on strong uptake for its Muse AI product. Offsetting the constructive tone, Seaport reiterated Nvidia at Sell due to capacity and competitive risks, and Bank of America downgraded Nike to Underperform on downside risk to EPS and valuation.

Analysis

The actionable read-through is a widening AI-infrastructure dispersion rather than a broad semiconductor beta trade. JCI’s data-center HVAC exposure can convert backlog into higher mix and operating leverage over the next 2-4 quarters, while TSEM and PDFS are more indirect beneficiaries of specialty-node capacity additions and yield-intensity. The key constraint remains advanced packaging and foundry allocation: if TSMC capacity stays tight, AMD can gain CPU ASP and enterprise-share economics without necessarily displacing NVDA’s accelerator revenue, making AMD/NVDA a cleaner relative-value expression than an outright NVDA short.

Seaport’s NVDA bear case is directionally useful but likely premature as a standalone catalyst: supply constraints generally preserve pricing and backlog visibility before they cap reported growth. The more credible 6-18 month risk is that hyperscaler internal silicon, AMD platform adoption, and customer resistance to bundled systems reduce NVDA’s incremental gross-margin and multiple upside. Watch NVDA data-center gross margin, networking attach rates, and disclosed supply commitments; sustained revenue growth above consensus with stable gross margin falsifies the short thesis.

JCI and STEM offer different versions of the power bottleneck trade. JCI monetizes construction and retrofit spending now, whereas STEM requires evidence that software revenue can overcome project-development cyclicality and financing sensitivity; the latter is a higher-beta earnings-revision setup, not yet a structural compounder. In consumer, the Nike downgrade matters more than the routine upgrades because classics weakness pressures both gross margin through promotions and wholesale shelf space, potentially creating a 2-3 quarter estimate-reset cycle that benefits performance-oriented competitors such as DECK and ONON.

Most rating initiations should produce only short-lived flow given limited fundamental novelty. The exception is Humana: an upgrade can matter if it coincides with improving medical-cost trends and 2027 Medicare Advantage rate visibility, but this remains a regulatory and utilization-risk trade rather than a valuation-only rebound.

Key Decisions for Investors

• Initiate a 3-6 month long JCI / short CARR pair: JCI has greater data-center cooling and backlog conversion torque, while CARR remains more exposed to normalized residential/light-commercial demand. Target 10-15% relative return; exit if JCI book-to-bill weakens below 1x or margin guidance fails to rise.
• Use a 6-12 month long AMD / short NVDA pair rather than outright semiconductor exposure. Size modestly because NVDA supply scarcity can extend its upside; add only if AMD server-CPU share or AI-platform design-win disclosures improve. Cover the NVDA leg if its data-center revenue and gross margin both exceed consensus for two consecutive quarters.
• Maintain or initiate a tactical 3-6 month short NKE versus long DECK or ONON. The thesis is negative estimate revisions from discounting and channel-share loss, not a one-quarter sales miss; take profits if Nike stabilizes gross margin and North America wholesale orders inflect positively.
• Treat STEM as an earnings-event watch rather than a full position until recurring software ARR, adjusted EBITDA, and operating cash flow each demonstrate sequential durability. If those metrics confirm at the next report, buy 6-month upside exposure with a defined stop below the pre-earnings low; failure on cash conversion invalidates the inflection narrative.
• Avoid acting on the NVDA sell call alone. Set an alert for evidence of falling accelerator pricing, lower networking attach, or a TSMC/advanced-packaging supply release; absent one of these, capacity tightness remains more supportive of near-term earnings than destructive.