Broadcom stock rises on Google, Anthropic deals

Broadcom announced a long-term agreement to develop and supply custom TPUs to Google and a supply-assurance deal to use Broadcom networking components in Google’s next-gen AI server racks through 2031, and expanded access for Anthropic to 3.5 GW of Google’s next-gen TPUs; Broadcom stock rose >3% on the news. Anthropic reported its revenue run-rate surpassed $30B (up from $9B at end-2025) and the number of customers spending >$1M annually doubled from 500 in February to 1,000 as of April 6. Analysts at BofA and Bernstein framed the deals as clearing insourcing/CoT concerns and improving demand visibility, supporting expectations of Broadcom gaining accelerator share in CY26–CY27; the stock is +110% over 12 months but down >6% YTD.

Analysis

The structural implication is that a supplier who can sell both purpose-built accelerators and networking components into the same rack captures a disproportionate share of wallet and bargaining leverage versus single-product vendors. That creates per-rack margin expansion opportunities driven by higher ASPs on integrated subsystems, lower logistics and validation cost for hyperscalers, and longer, more predictable revenue run-rates — effects that compound over 12–36 months as racks are refreshed and capacity is staffed.

Second-order winners are firms that supply complementary rack-level firmware, thermal and power-management IP (fewer vendors can undercut a combined silicon+networking bundle), while pure-play accelerator foundries and GPU-centric suppliers face incremental pricing pressure or forced specialization. Conversely, the move increases the value of captive tooling for hyperscalers that still prioritize control, which keeps a non-linear threat of insourcing alive; one large customer opting to internalize design tooling could wipe a multi-quarter growth runway.

Key risks split by horizon: over the next 90 days, narrative-driven position adjustments and option-gamma can create sharp volatility; across 6–24 months, execution on complex rack integrations, customer price negotiations, or a sudden slowdown in big-LM commercial spend are credible reversal catalysts. The consensus is rightly bullish on demand, but underappreciates the timing risk — meaningful margin capture requires visible share gains on installed racks and published supply commitments, not just commercial headlines. Monitor disclosed GW-equivalent commitments, network ASIC attach rates, and customer CAPEX cadence as the three highest-leverage datapoints.

Key Decisions for Investors

• Initiate an overweight in AVGO (size 3-5% net exposure) with a 12–24 month horizon — thesis: wallet-share + margin expansion. Set tactical stop at -10% from entry and trim 30% at +25% to realize rotation profits while leaving remainder to run on fundamentals.
• Defined-risk option trade: buy AVGO 12-month ATM call / sell 20% OTM call (ratio 1:1) sized to 1–2% of portfolio notional. Rationale: capture majority of upside if integration/attach proves out while limiting premium spend; target gross return 2.5x if AVGO rallies 20–30%.
• Relative-value pair: long AVGO / short NVDA notional 0.6x (delta-adjusted) over 6–12 months to isolate share-shift in custom accelerator wallet. This reduces market beta and profits if Broadcom gains incremental share or benefits from higher networking attach without broad GPU re-rating; take profits at 15–25% pair divergence, stop at 8–10%.
• Tail hedge: buy 9–12 month puts on GOOG/GOOGL 15% OTM to protect against a sudden hyperscaler CAPEX slowdown or pricing renegotiation that would compress demand across the stack. Cost is insurance against a concentrated downside event; treat as portfolio insurance sized to offset 20–30% of AVGO position gamma risk.