One Wall Street Analyst Says This Stock Is the Best Way to Invest in Anthropic and OpenAI. He's Right.

D.A. Davidson upgraded Broadcom to buy and raised its 12-month price target from $350 to $400, arguing the shares undervalue expected AI growth and future business with Anthropic and OpenAI. Broadcom projects AI revenue will double to $115 billion in fiscal 2027 and double again to $230 billion in fiscal 2028; Anthropic has $161 billion in commitments with Broadcom and plans to deploy 5 gigawatts of its co-developed TPUs in fiscal 2027 and another 10 gigawatts in fiscal 2028. OpenAI is expected to deploy 5 gigawatts of the chips in fiscal 2028. The article presents Broadcom as a lower-valuation way to gain exposure to the AI labs, while noting its ties to them have contributed to recent share-price weakness.

Analysis

The key underwriting question is not whether AI labs keep spending, but whether stated chip commitments convert into funded, deliverable deployments on schedule. Commitments are not equivalent to recognized revenue: financing access, power availability, deployment delays, and Broadcom’s ability to deliver can all push the payoff beyond the valuation window. Customer concentration also cuts both ways—successful deployments could establish Broadcom’s custom-chip platform, while a lab’s funding or product setback could create a sharp order and sentiment air pocket.

The strategic read-through is not automatically negative for NVIDIA (NVDA). Custom ASICs can take workloads where scale and efficiency justify bespoke silicon, while leaving demand for general-purpose accelerators intact; watch for evidence of workload substitution rather than treating announced ASIC capacity as a direct NVDA revenue loss. Alphabet (GOOG) and Meta (META) add customer diversity, but the article gives no quantified revenue contribution or independent confirmation of the lab commitments.

The valuation argument is fragile: the cited 2027 and 2028 multiples rely on forward adjusted-EPS estimates and exceptionally rapid AI growth. A modest schedule slip or lower conversion of commitments can make the apparent discount disappear without any collapse in long-run AI demand. Near term, positioning and updates to deployment plans matter more than IPO headlines; over 1–3 months, verify customer funding, purchase commitments, and Broadcom guidance. Over 6–18 months, successful deployments could validate ASIC economics and broaden the customer base. No trade should rest on the claim that IPOs alone de-risk the supply chain.

Key Decisions for Investors

• Consider a staged, modest long in Broadcom (AVGO), rather than buying aggressively on the valuation claim alone. Add only if subsequent disclosures confirm funded orders and deployment milestones; reassess if Broadcom reduces AI revenue guidance or customer deployments slip materially.
• Treat Anthropic and OpenAI IPO progress as a financing catalyst, not proof of chip demand. Track prospectus disclosures, cash needs, capital raised, and whether stated Broadcom commitments become binding orders; absent those details, keep the exposure on a watchlist.
• Do not short NVIDIA (NVDA) solely on ASIC announcements. Monitor evidence of workload migration, accelerator mix, and customer capex: sustained custom-chip adoption alongside resilient total spending supports both suppliers, while falling aggregate AI capex is the clearer sector risk.
• Before underwriting the headline valuation discount, reconcile Broadcom’s adjusted-EPS assumptions across fiscal 2027 and 2028 with reported AI revenue, delivery capacity, and cash conversion. Failure to meet milestones or a guidance cut would falsify the near-term long thesis.