Tech stocks today: Amazon to acquire satellite company Globalstar, OpenAI memo lashes out at rival Anthropic

Amazon agreed to acquire satellite company Globalstar for $11.57 billion, a major strategic move aimed at challenging SpaceX’s Starlink and expanding Amazon’s satellite ambitions to 3,200 satellites over three years. In AI, TSMC reported Q1 revenue up 35% on continued demand, while OpenAI and Anthropic exchanged sharp competitive critiques and Anthropic announced new multi-year cloud and chip-related agreements. Broader tech sentiment was helped by easing Iran-related geopolitical तनाव, even as the Middle East conflict has weighed on the sector for weeks.

Analysis

The cleanest read-through is that AI capex is becoming more concentrated in infrastructure owners while model-layer economics remain contested. That is structurally favorable for TSM, NVDA, AVGO, CRWV, and BE because the market is still underestimating how much incremental spend shifts from training to power delivery, networking, and custom silicon as hyperscalers try to de-risk shortages. The second-order loser is any AI software name whose valuation depends on durable gross-margin expansion without proving pricing power; the memo war reinforces that enterprise buyers are starting to demand trust, not just novelty.

Amazon’s satellite move is less about consumer broadband near-term and more about strategic insulation for its own cloud and logistics stack. If it can bundle connectivity with AWS and enterprise contracts, the real pressure falls on SpaceX-style vertical integration economics and on telecom backhaul providers that were counting on satellite as a niche adjunct rather than a platform layer. GSAT is the obvious headline winner, but the larger optionality sits in AMZN if this becomes a low-margin distribution channel that supports higher AWS retention and edge-compute penetration.

Oracle’s rally on power access is a reminder that the bottleneck in AI is migrating from chips to electrons. That should keep repricing power-adjacent assets with long-duration contracted cash flows, but it also raises execution risk: if grid interconnection, transmission, or permitting slip, the market will eventually punish forward capex promises. The geopolitical backdrop matters too: any renewed Middle East escalation would likely hit high-beta software and semis first through multiple compression, not earnings revisions, so the trade is faster than fundamentals.

The contrarian view is that the market may be over-discounting the intensity of AI demand while under-discounting regulation and safety backlash. The internal AI feud is noise tactically, but it signals a coming phase where procurement committees care more about governance, uptime, and auditability than raw benchmark wins. That usually favors incumbent platforms with distribution and capital intensity, and it argues for owning the picks-and-shovels rather than assuming the current model leaders win all the economics.

Key Decisions for Investors

• Long TSM / long NVDA into the April earnings cycle; use any 3%-5% pullback to add. Risk/reward remains skewed to the upside as semiconductor demand is still outrunning supply and consensus underestimates 2026 capex duration.
• Pair long CRWV vs short a basket of AI software names with weak monetization visibility (e.g., high-multiple pure plays). Thesis: infrastructure monetizes first while software faces procurement scrutiny and margin pressure over the next 3-6 months.
• Buy AMZN on a 1-2 week horizon, but express as call spreads rather than stock. The satellite transaction is a strategic option on AWS distribution and edge connectivity; downside is limited if the deal is mostly infrastructure-financing, upside could re-rate if it is tied to enterprise bundle sales.
• Add to BE or related power-enablement exposure on weakness over the next month. The market is beginning to price power as a scarce AI input; contracts with hyperscalers can re-rate these names if backlog turns into recurring revenue.
• Short-term hedge: buy NVDA or SOXX downside puts into any geopolitical escalation spikes. The setup is vulnerable to multiple compression over days, even if 2-4 quarter fundamentals remain intact.