Built by Claude Fable

AKAM | Concerns, Catalysts & Risks: 6/10

Deep dive pages: Overview | Business Model | Financial Trends | Thematic Exposure | Management Quality | Investor Sentiment | Concerns, Catalysts & Risks

Bottom line: AKAM trades below both its mature-infrastructure peers and the neoclouds on EBITDA (about 10x FY2027 EV/EBITDA vs a peer average of about 19x), but not on cash flow: FY2026 FCF turns negative, so P/FCF is not meaningful. It also has the densest near-term catalyst calendar it has had in a decade: the Q3 print on 2026-11-05, the CIS ramp in Q4 2026, the low-teens growth guide for 2027, and the Anthropic revenue start in 2H27. Direct China revenue is not disclosed and is likely immaterial. The China-adjacent exposure is the ByteDance/TikTok account (about 6% of revenue at the February 2025 disclosure, and shrinking as that customer builds its own delivery). Two things hold the score at 6. First, the balance sheet and FCF now carry real execution risk: about $3.5B of FY2026 capex and ~$5.5B for the Anthropic build, FCF falling while revenue grows, and $7.64B of convertible notes (about $9.4B including leases). Second, the company is becoming concentrated on one counterparty whose $11.6B commitment has delivery-based termination outs.

1. Valuation table (primary metric: EV/EBITDA, FY+1 = FY2027)

AKAM is a mature tech franchise in the middle of a capex-heavy pivot to compute, so EV/EBITDA is the primary metric and P/E is a cross-check. FY+1 is FY2027 because FY2026 ends in about 10 weeks. Market data comes from the FMP /stable quote, key-metrics-ttm and analyst-estimates endpoints, pulled 2026-10-19. Peer multiples are computed the same way: EV (FMP TTM) divided by FMP FY2027 consensus. F5 uses FY Sep-2027 and IREN uses FY Jun-2027.

Peer detail (FY2027 consensus, FMP):

• AKAM is below the peer average on every EBITDA and earnings lens, but not on cash flow. Only CHKP (a no-growth security name) and CRWV (a levered GPU lessor) trade in line. That makes sense: the market prices AKAM as a declining CDN plus a security annuity, and gives little credit to a CIS business that just signed more than $14B of multiyear commitments.
• Caveat 1, EBITDA definition. FMP consensus EBITDA ($1,940M FY2026E) runs above Akamai's own adjusted EBITDA trajectory. FY2025 adjusted EBITDA was $1,801.6M at a 43% margin. Q2 2026 came in at $416.1M and a 38% margin, and Q3 is guided to 38% to 40%. On company-defined EBITDA the FY2027 multiple is probably 1x to 2x higher than shown.
• Caveat 2, EV. The FMP EV ($20.63B) embeds net debt of about $7.86B. A sourced June 30, 2026 bridge is lower: market cap $12.77B plus $7,562.8M of convertible notes (carrying, current plus long-term) plus $1,776.2M of operating leases ($370.4M + $1,405.7M), less $4,616.3M of cash and marketable securities ($1,480.3M + $1,875.1M + $1,260.9M), gives EV of about $17.5B, or about 8.6x FY2027 EBITDA.
• Caveat 3, cash flow. P/FCF is 20.3x on TTM FCF (FMP) but not meaningful for FY2026, when capex of about $3.5B exceeds operating cash flow. A lessor-style build should be judged on FCF, and on that lens AKAM is not cheap. Both caveats push in opposite directions, and neither changes the conclusion that AKAM trades below peers.

2. China exposure

Read: direct China revenue is below 10% and probably low single digits. The real China-linked risk is the ByteDance account, which is already in managed runoff inside Delivery.

3. Catalysts (timing and size)

Anthropic deal, verified terms (8-K): $11.6B contractual commitment over 7 years for Anthropic's CPU workloads. Warrant on non-voting convertible Series B preferred for 7.7M common shares as converted (up to about 5% of shares), exercise price $111.33, 7-year term. About 2% vests with the $11.6B commitment and about 1% more per additional $3B of purchases. No impact on 2026 revenue guidance.

Verified from the 8-K body (Item 1.01): the commitment is subject to delivery and service availability requirements; Anthropic can terminate a plan on a material outage (subject to conditions) or the MSA on a change of control in favor of a direct competitor; the MSA dates from May 5, 2026, so the May $1.8B deal is Plan 1 by date and additive to the $11.6B in Plans 2 and 3. Still not verified: the financing plan, and whether the $1.7B memory pre-buy counts toward the $5.5B build.

Sneaky neocloud math: CIS TTM revenue is $368.8M (sum of Q3 2025 $81.2M, Q4 2025 $93.7M, Q1 2026 $94.6M, Q2 2026 $99.3M). The Anthropic deal alone averages about $1.66B/yr ($11.6B / 7), which is 4.5x current CIS revenue and about 32% of FY2027 consensus total revenue. At the reported end-2028 $1.7B run rate, Anthropic would be roughly 27% of FY2028 consensus revenue ($6.37B). That is a neocloud-style revenue profile on a security and CDN multiple.

4. Concerns and red flags (trajectory first)

1. Revenue up, FCF down (classic red flag). FY2025 revenue grew 5.4% while FCF fell 16% to $699.3M. 1H26 FCF was $221M vs $291M in 1H25. FY2026 capex reconciles to about $3.5B: the August guide of about 40% of revenue (about $1.8B) plus the roughly $1.7B memory pre-buy added by the 8-K. That is against about $1.75B of adjusted EBITDA (our estimate from the guidance). FY2026 and likely FY2027 FCF turn sharply negative. The cause is deliberate (pre-revenue build), but the cash is out the door before the contracts pay.
2. Leverage is rising quickly. Debt was $6.91B at FY2025 year-end including leases (FMP). At June 30, 2026 it was $7,640.0M of convertible notes principal plus $1.78B of operating leases, about $9.4B, against $4.62B of cash and securities (Daloopa). Akamai added $3.5B of zero-coupon converts in May 2026, paused buybacks in Q2, and still has about $5.5B of Anthropic-related capex to fund. Net debt/EBITDA was 6.1x on a GAAP-EBITDA basis per FMP. More converts or other debt look likely, and any equity-linked financing adds to the warrant dilution.
3. Customer concentration flips from TikTok to Anthropic. One AI lab could be about 25% to 30% of revenue by 2028, and its commitment is conditional on Akamai hitting delivery and availability SLAs, with termination outs. If Anthropic's own compute economics or funding change, AKAM carries the stranded capex.
4. Margin compression is visible. Adj. EBITDA margin fell from 43% to 38% over four quarters and non-GAAP EPS declined 8% YoY in Q2. Management's deal framework (cash gross margin mid-60s to mid-70s, operating margin low-20s to low-30s) is below the legacy 30%+ operating margin, so mix dilutes margins even when the deals work.
5. Beat magnitude is shrinking. EPS beats narrowed from $0.13 to $0.22 across FY2025 to $0.01 to $0.02 in 1H26 (FMP). Q3 guidance also came in about 1.4% below consensus at the August print.
6. Delivery is structurally declining at mid single digits even with traffic growth, and management guides it down mid single digits in 2026.
7. Warrant dilution. 7.7M shares (about 5%) at $111.33. About 2% vests with the base commitment, so it is real dilution if the stock clears the strike.

5. Regulatory / political risk

• Moderate, not severe. The main risk factors are data localization and sovereignty (EU frameworks are expanding, US-EU tension is cited in the 10-Q), the DOJ countries-of-concern data rule, and U.S. export controls on advanced GPUs. Export controls matter because Akamai's distributed Inference Cloud deploys NVIDIA Blackwell outside the U.S. The TikTok/ByteDance account carries U.S. political risk, though it is already shrinking as a share of revenue.
• No antitrust overhang. Akamai is a share-taker against hyperscalers in compute and a large but not dominant security vendor.

6. 10x web/API traffic scenario (illustrative sensitivity, not a forecast)

FY2025 base: Delivery $1,256.7M, Security $2,243.4M, CIS $313.9M.

Illustrative sensitivity, not a forecast. This is the single version used on the Overview, Thematics and Valuation pages: a five-year horizon from FY2025, calibrated to Akamai's own history. Akamai does not disclose traffic, so revenue per unit of traffic cannot be measured directly. The anchor is that Delivery revenue fell 25% from FY2022 to FY2025 (-9%/yr) while traffic grew, so even 10x traffic is unlikely to lift Delivery; the Security cases bracket its own recent growth rates of 5% to 15% a year. Takeaway: a 10x traffic world does not rescue Delivery (history says volume does not offset deflation) and helps Security, which is already 55% of revenue and has less commodity pricing. The traffic thesis therefore leans on Security and CIS, not on the CDN. That supports the bull case without depending on it.

7. Bull vs bear

Bull case. Akamai has signed more than $14B of multiyear CIS commitments in 2026, and those commitments convert a no-growth CDN into a low-teens to 20%+ grower (FMP consensus FY2028 +22.4%) while the stock trades at about 10x FY2027 EBITDA, half the neocloud average. Its distributed CPU footprint is the scarce asset Anthropic chose for agentic workloads. The expansion option (up to $9B more) and a sold-out GPU fleet suggest the commitments are a floor. Security keeps compounding at about 10% and is a natural beneficiary of AI-driven attack surface.

Bear case. Akamai is becoming a levered capex lessor to one AI lab at worse margins than its legacy business, with FCF negative for two years and $7.64B of convertible notes (about $9.4B with leases). Anthropic can walk if SLAs slip. The low multiple reflects a correct haircut on capital intensity, and EV/EBITDA flatters a business whose EBITDA will be consumed by depreciation. Delivery keeps shrinking, beats keep narrowing, and the dilutive warrant goes to the customer with the most leverage over Akamai.

Scoring

Final score: 6/10.

Assessment: AKAM's risk/reward profile is unusual. Valuation and catalysts would score 8 to 9 on their own: the stock trades at about 10x FY2027 EV/EBITDA against a 19x peer average, and the 8-K-verified $11.6B, 7-year Anthropic CPU agreement plus about $2.8B of other 2026 CIS wins give CIS a clear multiyear ramp. Most of the ramp starts in Q4 2026 and 2H27, with Anthropic alone potentially reaching about a $1.7B annual run rate by end-2028, 4.5x today's entire CIS business. Direct China revenue is immaterial. The China-linked ByteDance account is in managed decline, and regulatory risk is moderate (data sovereignty, GPU export controls) with no antitrust overhang. What keeps the score at 6 is that the bull case is financed up front. Revenue is growing while FCF falls, FY2026 capex will roughly double adjusted EBITDA, gross debt is $7.64B of converts (about $9.4B with leases), margins are compressing, and by 2028 one counterparty could be a quarter of revenue under a commitment with delivery-based termination outs. That is neocloud execution risk at a legacy-CDN multiple. The discount is partly earned, and the Q3 print on 2026-11-05 (capex, financing plan, 2027 framework) is the first real test.

Sources: Daloopa (company_id 251; citations inline); FMP /stable quote, key-metrics-ttm, analyst-estimates (pulled 2026-10-19); Akamai earnings call transcripts FY2024Q3 to FY2026Q2; Akamai 8-K Ex. 99.1, 2026-09-24; Akamai press release; TechCrunch via Yahoo Finance; Bloomberg, 2026-05-08; Dow Jones via itiger on TikTok exposure; 10-Q 2026Q2 risk factors.

Data sourced from Daloopa