Tech stocks today: Amazon to acquire satellite company Globalstar as it takes on Elon Musk's Starlink

Tech stocks rose as Iran de-escalation hopes improved risk sentiment, while Amazon agreed to acquire Globalstar for $11.57 billion, a move seen as a direct challenge to SpaceX's Starlink. OpenAI and Anthropic intensified their rivalry, with OpenAI's revenue chief calling Anthropic's $30 billion run-rate "inflated" and Anthropic announcing a multi-year CoreWeave infrastructure deal while exploring its own chips. Separately, TSMC said Q1 revenue rose 35% on AI demand, and Oracle gained 7% after expanding its power agreement with Bloom Energy.

Analysis

The cleanest read-through is not “AI is hot” but that the capex stack is still widening, which keeps second-order winners intact even if model-level competition gets noisier. Anthropic’s apparent need for more external compute while simultaneously exploring custom silicon reinforces a multi-year bottleneck: frontier labs are being forced to pre-commit demand across clouds, chip designers, and data-center operators earlier in the product cycle. That favors infrastructure names with contracted capacity and penalizes any AI story reliant on flexible access to compute at spot pricing.

Amazon’s satellite move is strategically more interesting as a distribution and routing play than as a telecom headline. If it forces a price war or a subsidy race against Starlink, the near-term winner is the broader launch and ground-segment supply chain, while the loser may be the economics of low-margin consumer broadband on both sides. The real option value for AMZN is not satellites per se, but tighter integration of connectivity into AWS logistics, remote industrial customers, and defense-adjacent workloads over a 2-4 year horizon.

The market seems to be underweighting the monetization power of AI infrastructure bottlenecks. Oracle’s power agreement and TSMC’s revenue growth both point to the same trade: power, chips, and specialized cloud capacity remain the scarce inputs, so pricing power accrues to the “picks and shovels” layer before it reaches application software. That argues for staying long the infrastructure complex while being more selective on software names that depend on broad enterprise adoption rather than exclusive distribution or hard-to-replicate compute access.

Contrarian risk: if AI capex keeps accelerating, investors may eventually treat it like a utility cycle and start discounting diminishing marginal returns, especially if enterprise demand doesn’t broaden beyond a handful of use cases. In that scenario, the market could rotate from “all AI infrastructure up” to a narrower winner set within 1-2 quarters. Geopolitically, any durable de-escalation in the Middle East would mostly remove a discount on tech multiples, which is supportive but likely less important than the ongoing supply-side AI race.

Key Decisions for Investors

• Go long NVDA/AVGO/TSM as the core AI bottleneck basket for the next 3-6 months; risk/reward remains favorable while compute scarcity persists, but trim on any evidence of cloud buyers delaying orders.
• Pair long CRWV vs short a basket of lower-conviction AI software names over 1-2 quarters; external capacity providers should monetize the current shortage before margin pressure normalizes.
• Buy AMZN on weakness and hold 6-18 months; the satellite move is a strategic optionality enhancer, and downside is limited unless integration costs or regulatory scrutiny rise sharply.
• Avoid chasing ORCL after the move; use strength to fade if the stock rerates beyond infrastructure peers, since the market may be pricing a one-time power/capex catalyst rather than durable acceleration.
• For a tactical hedge, short GSAT against long AMZN into deal completion volatility; capture takeover spread while limiting exposure to standalone satellite execution risk.