Raymond James maintains Outperform on Akamai stock after Anthropic deal

Raymond James reiterated an Outperform rating and $170 price target for Akamai after its expanded seven-year, $11.6 billion Anthropic AI-infrastructure agreement, which implies roughly $1.66 billion in annual recurring revenue. The firm expects the deal to lift compute and infrastructure-services EBITDA margins into the mid-50% range by 2028 and company non-GAAP operating margins into the low 30% range, although material returns are delayed until 2028. The agreement produces an estimated $0.31 of revenue per $1 of capex with an approximately four-year payback period, supporting ROIC in the high-single-digit to low-double-digit range.

Analysis

The key underwriting issue is not contracted revenue but the spread between project ROIC and Akamai’s cost of capital. A high-single- to low-double-digit return profile leaves little cushion once financing costs, accelerated hardware depreciation, power inflation, and customer-specific buildout risk are included; this is a lower-quality dollar of growth than the legacy edge-security model. A long-duration fixed-price commitment can therefore expand reported revenue while diluting consolidated free-cash-flow conversion through 2027.

Anthropic concentration creates a financing and counterparty-risk loop that the headline backlog figure obscures. If Anthropic raises capital at a lower valuation, moderates training spend, or shifts workloads to strategic cloud partners, Akamai could retain nominal backlog but face renegotiation, delayed deployment, or underutilized capacity; the relevant disclosure is cancellation protection, prepayment, minimum-utilization terms, and who owns residual hardware. The market will likely reward evidence of customer-funded capex and lease-like economics, not analyst-derived 2028 margins.

Near term, the setup is mixed: estimate revisions can support AKAM over the next 1-3 months, but the capex-to-cash-flow gap makes this vulnerable to any increase in funding needs or weaker legacy Security/Delivery growth. Over 6-18 months, AKAM is competing for AI workloads against ORCL, CRWV, NBIS, MSFT and GOOGL, whose balance sheets, GPU procurement scale, and embedded customer relationships can compress pricing on follow-on capacity. The contrarian view is that the share-price weakness is rational unless management demonstrates returns materially above WACC rather than merely validating a large contract.

Key Decisions for Investors

• Maintain AKAM as a watch-list long rather than chase the initial analyst-target dispersion; initiate only after quarterly disclosures show capex funding, committed capacity, and free-cash-flow guidance consistent with project returns above 12%. A guidance cut to FCF or incremental debt/equity funding would falsify the thesis.
• For a 1-3 month relative-value expression, consider long AKAM / short ORCL only if AKAM trades below its pre-contract valuation multiple while ORCL rerates on AI infrastructure enthusiasm; size modestly because ORCL’s broader cloud earnings can overwhelm the pair. Exit if AKAM signals a material capex increase without matching minimum-revenue commitments.
• Avoid treating the contract as a 2026-27 EPS catalyst. Use post-earnings weakness to assess long-dated exposure only after management quantifies depreciation, power costs, and customer prepayment; absent those data, the risk/reward is asymmetric toward multiple compression.
• Monitor Anthropic financing, cloud-partner announcements, and AKAM backlog conversion each quarter. Any Anthropic workload expansion with third-party hyperscalers or a backlog-conversion miss is a trigger to underweight AKAM, as it would challenge the assumed utilization curve behind the margin ramp.