Stock Movers: Akamai, Oracle, Everpure (Podcast)

Anthropic signed an $11.6 billion, seven-year computing contract with Akamai, the largest deal in Akamai's history, requiring an estimated $5.5 billion in associated capital spending. Oracle issued a force-majeure notice related to the Project Jupiter New Mexico data center, potentially deferring rent if power-related commitments delay the facility. Everpure shares surged as much as 20% after its 2028 revenue forecast exceeded average analyst estimates.

Analysis

AKAM’s rerating should be constrained by the financing math rather than headline contract value. The incremental build requires roughly $5.5B of capital against revenue recognized over seven years, while the embedded warrant transfers part of the upside to the customer; investors should underwrite utilization, power pass-throughs, depreciation and debt-funded interest expense before assigning hyperscaler-like multiples. The non-obvious read is that a CPU-focused deployment could be lower-margin and more substitutable than scarce GPU capacity, leaving AKAM exposed if Anthropic changes model architecture or secures cheaper capacity from cloud incumbents over the next 12-24 months.

ORCL’s project issue is more important as a signal of grid-delivery risk than as a single-site rent dispute. If power availability is gating commissioned capacity, Oracle’s booked AI infrastructure demand may convert into revenue and cash flow later than investors expect, while fixed commitments and construction-related costs remain. This creates a potential spillover to data-center financing vehicles, including OWL-linked strategies: delayed energization can extend construction periods, defer lease cash flows and increase the refinancing burden on assets financed against aggressive completion assumptions over the next 1-3 quarters.

The consensus is likely treating AI compute demand as the scarce resource; increasingly, deliverable power and completed interconnection are the scarce inputs. That favors incumbent operators with energized campuses and contracted generation, but penalizes developers whose valuation assumes on-time capacity delivery. The move in P should not be chased without reconciling the reported business description, ticker identity and the durability of its 2028 framework; a long-duration target has limited near-term valuation support absent intermediate margin, backlog and free-cash-flow milestones.

Key Decisions for Investors

• Do not chase AKAM on the initial announcement; establish a 1-3 month watch for disclosed funding mix, expected annual capex cadence, customer concentration and incremental EBITDA margin. Consider a tactical long only if management demonstrates that incremental returns exceed its cost of capital without material net-leverage expansion; exit on reduced utilization assumptions or a material increase in capex guidance.
• Express the power-delay thesis via a 3-6 month ORCL underweight versus MSFT rather than an outright short. MSFT has greater existing cloud scale and less dependence on any single delayed campus; cover if Oracle confirms Project Jupiter energization, reiterates near-term cloud revenue guidance, or if the relative spread widens materially without additional capacity-delay evidence.
• Reduce or hedge OWL exposure around disclosures on data-center fund duration, construction-loan marks and lease commencement assumptions. The key falsifier is evidence that delayed facilities retain contracted economics with no material extension of financing duration or impairment to fee-related earnings.
• Treat P as an alert rather than a trade until ticker/entity verification and the bridge from current results to 2028 revenue are available. A credible entry requires quarterly evidence of backlog conversion and margin progression, not merely an out-year target.