Built by Claude FableAKAM | Investor Sentiment: 7/10
Deep dive pages: Overview | Business Model | Financial Trends | Thematic Exposure | Management Quality | Investor Sentiment | Concerns, Catalysts & Risks
Inverted scoring. The question is whether management is specifically and repeatedly bullish on something the street does not believe. Being down or "hated" does not count as a signal.
Bottom line: There is a real, specific divergence. Management has spent four years saying distributed compute (Linode, then Cloud Infrastructure Services, now Inference Cloud) would become Akamai's main growth engine. It now has about $14.4B of signed multi-year CIS commitments to show for it. Management also says these contracts earn low-20s to low-30s non-GAAP operating margins and produce "phenomenal" free cash flow once the capacity is built. The street accepts the revenue but not the economics. Consensus EPS is flat from FY2026 to FY2027, ratings split 26 Buy to 26 Hold/Sell, and the downgrades all point to CIS margins, capex intensity and Anthropic concentration. The thesis is not that analysts ignore CIS. It is that they model the bookings as low-return capex, while management says the contracts earn a return. Insider confirmation is weak: the CEO's $6.6M of buying ($6.8M with the Chair) happened in 2025, before the deals, and Form 4s show $17.1M of open-market sales by 11 insiders and no purchases in the 12 months to 2026-10-19. That keeps the score from reaching 8 or above.
4.1 Management-Street Divergence (primary, 50% of score)
What management keeps saying (specific, repeated, quantified)
| Call | Management claim | Source |
|---|---|---|
| FY2024Q4 (Feb 2025) | Goal of ~20% compute revenue CAGR over 3 to 5 years; CIS ARR growth 40% to 45% in 2025 | transcripts/AKAM_FY2024Q4.txt |
| FY2025Q4 (Feb 2026) | CIS growth to accelerate to 45% to 50% in 2026, driven by AI Inference Cloud | transcripts/AKAM_FY2025Q4.txt |
| FY2026Q1 (May 2026) | $1.8B 7-year "frontier model company" deal, largest in company history; total revenue growth to reach double digits in 2027; GPU pipeline "significantly exceeds" inventory | transcripts/AKAM_FY2026Q1.txt |
| FY2026Q2 (Aug 2026) | $600M 4-year robotics GPU deal; >$2.8B YTD commitments; 2027 growth "low teens"; GPU capacity "completely sold out"; large deals earn mid-60s to mid-70s cash gross margin and low/mid-20s to low-30s non-GAAP operating margin; buyback paused because "we have not seen a growth opportunity like this in a very, very long time" | transcripts/AKAM_FY2026Q2.txt |
| 8-K, 2026-09-24 | Anthropic Project Plans 2 and 3: ~$11.6B over 7 years of dedicated, mainly CPU, capacity; up to $9B more possible (~$20B total); ~$5.5B capex; 2026 capex up ~$1.7B for memory pre-buy; warrant for up to 7.74M as-converted shares at $111.33 | 8-K, Ex. 99.1 |
Anthropic deal, verified terms vs unverified
- Verified (8-K): Master Services Agreement with Anthropic dated May 5, 2026. Project Plans 2 and 3 signed Sept 18, 2026, totaling ~$11.6B, with a 7-year term from each plan's service start date. Service start dates are not disclosed. Anthropic can terminate a plan on a material outage, subject to conditions. Warrant: 387,051 Series B preferred shares (20:1, so 7,741,020 common), cash exercise only. 40% vests on first payment under Plan 3, then 20% per additional $3.0B of committed value. Jabil was authorized to buy ~$1.7B of memory components. A Lenovo hardware SOW runs 7 years.
- Confirmed counterparty per the 8-K: Anthropic has been the customer under the MSA since May 5, 2026, and the May 2026 $1.8B "frontier model" deal matches Project Plan 1 by date (Bloomberg reported Anthropic at the time, Yahoo/Bloomberg), and the 8-K does not size Plan 1). The press release mentions ">$2.8B from other customers", which is ambiguous on this point.
- Revenue timing: no 2026 impact from the $11.6B. Morgan Stanley sees $150M to $300M of 2027 revenue from it, and third-party reports put the start in H2 2027 (Yahoo/Stocktwits). Plan 1 adds ~$20M to $25M in Q4 2026 (FY2026Q1 call).
Contracted run-rate vs what CIS is today (rough, straight-line, full-ramp)
| Contract | Value | Term | Implied annual revenue at full ramp |
|---|---|---|---|
| Anthropic Plans 2+3 (8-K) | $11.6B | 7 yrs | ~$1.66B |
| Frontier model deal, likely Anthropic Plan 1 | $1.8B | 7 yrs | ~$0.26B |
| Robotics GPU deal (Q2 call) | $0.6B | 4 yrs | ~$0.15B |
| US tech deal (Feb 2026) | $0.2B | 4 yrs | ~$0.05B |
| Total | ~$14.2B | ~$2.1B/yr |
FY2025 CIS revenue was ~$314M ($67.6M + $71.5M + $81.2M + $93.7M). At full ramp the signed contracts add roughly 6.7x FY2025 CIS revenue and about 47% of FY2025 total revenue. DA Davidson counts $14.4B of total CIS commitments YTD and estimates ~$4 of added EPS by 2029 (GuruFocus).
Daloopa trajectory: management's narrative vs reported numbers
| Metric | 2024Q3 | 2024Q4 | 2025Q1 | 2025Q2 | 2025Q3 | 2025Q4 | 2026Q1 | 2026Q2 |
|---|---|---|---|---|---|---|---|---|
| Total revenue | $1,004.7M | $1,019.9M | $1,015.1M | $1,043.5M | $1,054.6M | $1,094.9M | $1,073.6M | $1,099.7M |
| Total revenue YoY | +4.1% | +2.5% | +2.9% | +6.5% | +5.0% | +7.4% | +5.8% | +5.4% |
| CIS revenue | $58.3M | $64.7M | $67.6M | $71.5M | $81.2M | $93.7M | $94.6M | $99.3M |
| CIS YoY | 27% | 35% | 30% | 30% | 39% | 45% | 40% | 39% |
| Security revenue | $518.7M | $534.6M | $530.7M | $551.9M | $568.4M | $592.4M | $589.8M | $604.4M |
| Security YoY | 14% | 14% | 8% | 11% | 10% | 11% | 11% | 10% |
| Delivery + other cloud apps | $427.7M | $420.7M | $416.8M | $420.1M | $405.0M | $408.9M | $389.2M | $395.9M |
| RPO | $3.5B | $4.3B | $4.3B | $4.3B | $4.8B | $5.2B | $5.5B | $7.6B |
| RPO YoY | +9% | +26% | +26% | +26% | +37% | +21% | +28% | +77% |
| cRPO ($B, derived) | 2.28 | 2.58 | 2.58 | 2.58 | 2.64 | 2.86 | 2.75 | 3.04 |
| cRPO YoY (derived) | +9% | +17% | +17% | +17% | +16% | +11% | +7% | +18% |
| Capex | $164.0M | $192.7M | $225.9M | $214.2M | $224.2M | $153.7M | $206.3M | $346.5M |
| Capex YoY | n/a | n/a | n/a | n/a | +36.7% | -20.2% | -8.7% | +61.8% |
| Capex % revenue | 16% | 19% | 22% | 21% | 21% | 14% | 19% | 32% |
Why the trajectory matters: CIS growth accelerated from 30% to 45% through 2025, then stalled at ~39% to 40% in 1H26 while capex roughly doubled. The street sees spending today and revenue later. RPO jumped $2.1B in Q2 (the $1.8B deal) to $7.6B, +77% YoY, while derived cRPO grew only about 18% because the new contracts are long-dated. RPO does not yet include the $11.6B Anthropic plans, so the Nov 5 print should show the first hard evidence. Management's "big hockey stick" for CIS in Q4 2026 (Q2 call) is the near-term test.
Where the street pushes back (Q&A, FY2026Q1 and FY2026Q2)
| Theme | Analysts | What they asked | Management answer |
|---|---|---|---|
| Deal economics / margin | DiFucci (Guggenheim), Ader (KeyBanc), Power (Baird) | Will margins fall structurally? When does profit arrive? Is component inflation passed through? | Cost-plus style repricing mechanisms; full profitability within about a quarter of racking; low-20s to low-30s operating margin |
| Funding / balance sheet | Singh (MS), Boolani (Citi) | Net debt or equity to fund the pipeline? | $3.5B zero-coupon converts, buyback paused, IG rating kept, "no equity offerings" so far |
| CIS ramp credibility | Singh (MS) | Two quarters of sub-40% CIS growth vs a ≥50% FY target: what gives confidence in the Q4 hockey stick? | GPU delivery slipped from Q2 to Q3; big contracts start Q4 and ramp through 2027 |
| Distributed vs neocloud | Colville (Scotia), Kessinger (DA Davidson), Boyd (UBS) | Is the mega deal just a few central data centers, i.e. a neocloud? What edge does Akamai have? | Will not discuss the deal; 700 cities, 4,300 locations, lower egress cost, backbone flexibility |
| Power / capacity | Fish (Piper), Louthan (Raymond James), Van Rhee (Craig-Hallum) | MW available, sites needed, community pushback | 5 to 10 MW core sites expandable to 20 to 30 MW; "not concerned" about power or colo |
Read: Q&A tone is congratulatory on demand. Nobody disputes the bookings. The doubt is entirely about returns, funding and execution. Outside the calls, the doubt is explicit:
- HSBC cut to Hold (PT $171 to $123) on weaker CIS margins, ~8.5% EPS CAGR through 2028, and capex averaging ~36% of revenue (Yahoo).
- An unnamed Oct 8 downgrade cited CIS margins and multi-year capex (Quiver).
- DA Davidson cut its PT to $150 from $185 on capital intensity while keeping Buy.
- Independent work models Anthropic project IRRs of 7.4% to 10.7%, roughly value-neutral against a 10% hurdle at a 25% margin, with Anthropic at ~24% of 2029 revenue (Seeking Alpha).
Management outlook vs consensus (FMP analyst-estimates, 2026-10-19)
| Item | FY2026E | FY2027E | FY2028E |
|---|---|---|---|
| Management | Rev $4.445B to $4.530B (low, high); EPS $6.40 to $7.05; CIS ≥50% | "Low teens" total growth (stated Aug, BEFORE the $11.6B deal) | Not guided; contracts imply ~$2.1B/yr CIS run-rate at full ramp |
| Consensus revenue | $4,490M (+6.7%) | $5,209M (+16.0%), range $4,877M to $5,399M | $6,374M (+22.4%), range $5,331M to $7,113M |
| Consensus EPS | $6.68 | $6.74 (flat) | $8.61, range $5.96 to $12.08 |
What this shows: on revenue, the street has already moved above management's pre-Anthropic "low teens" 2027 call. So the divergence is not about top-line disbelief. It is about earnings power. Management says these deals carry corporate-average or better operating margins and strong post-build FCF. Consensus still has FY2027 EPS flat on 16% revenue growth, and the FY2028 EPS range is 2x wide ($5.96 to $12.08). That spread is the street saying it does not know whether this is a utility-return capex program or a high-return platform. The 10x-traffic angle feeds this too. Leighton said (Q1 call) agent traffic "does not make a huge swing in bits delivered" unless it is video. The monetization path he describes is compute first, security second (WAF, bot/agent management, API security, LayerX). Analysts keep modeling Delivery as a melting ice cube at -5% to -7%, which is consistent with management, so there is no divergence there.
Divergence verdict: specific and repeated, but on returns, not demand. It scores well under the NVDA template, where management says something positive and the street does not believe it, but it is less clean than a case where analysts dismiss the growth driver outright.
4.2 Analyst consensus
| Source | Buy | Hold | Sell | Avg PT | Notes |
|---|---|---|---|---|---|
| FMP grades-consensus | 26 (1 SB + 25 B) | 25 | 1 | $167.27 (median $170, range $123 to $225) | PT set includes stale May/Sept targets |
| MarketBeat (Oct 17, 2026) | n/a | consensus "Hold" | n/a | $150.10 (24 analysts) | per web search |
| Recent actions | MS OW $165, DA Davidson Buy $150 (cut from $185), Oppenheimer OP $180, Guggenheim Buy $225 | UBS Neutral $148, RBC SP $135, HSBC Hold $123, JPM Neutral (upgraded from UW in Aug) |
Current price $88.845, market cap $12.77B (FMP quote, 2026-10-19). Every published target sits well above the current price, even the low of $123. That fits a split, slow-to-update street, not a crowded long. Ratings are close to 50/50 Buy vs Hold, which is "mixed/skeptical", the favorable end of the rubric.
4.3 Insider activity
| Period | Insider(s) | Form 4 code | Amount | Signal |
|---|---|---|---|---|
| 2025-02-27 | CEO Tom Leighton | P (open-market buy) | $3.0M (37,670 sh) | Strong, before the CIS deals |
| 2025-08-11 | CEO Tom Leighton | P | $3.6M (50,000 sh at ~$72) | Strong |
| 2025-08-11 | Chair Daniel Hesse | P | $0.2M (3,000 sh) | Confirming |
| 2025-10-19 to 2026-10-19 | 6 insiders: Sundaram $3.2M, Joseph $2.1M, CFO McGowan $2.0M, CTO Blumofe $1.3M, Karon $0.8M, Salem-Jackson $0.6M | S under 10b5-1 plans | $9.9M | Mildly negative, pre-planned |
| 2025-10-19 to 2026-10-19 | 6 insiders: Karon $1.8M, Williams $1.6M, GC Ahola $1.2M, CAO Howell $1.1M, directors Miller $1.1M and Ranganathan $0.3M | S, no 10b5-1 flag | $7.2M | Negative, discretionary |
| 2025-10-19 to 2026-10-19 | CEO ($6.3M) and 9 other officers | F (tax withholding on vesting) | $19.0M | Neutral, not a sale decision; excluded |
| 2025-10-19 to 2026-10-19 | All insiders | P | $0 | No buying |
Read: the CEO put $6.6M of his own money in during 2025 ($6.8M including the Chair), near the lows and before the CIS bookings. That is a real track-record signal. Since then the flow has run the other way: $17.1M of open-market sales by 11 insiders in the last 12 months, $7.2M of it outside 10b5-1 plans, and no purchases. This is the single reconciled figure used on every page (Form 4 codes from SEC EDGAR). The earlier read that the CEO and CFO had only vest and tax-withholding dispositions was wrong for the CFO, who sold $2.0M under a 10b5-1 plan. The Sept 30 quarter-end to Nov 5 earnings blackout likely explains part of the recent gap, so watch for a post-print Form 4.
4.4 Retail / social sentiment
- At the Sept 25 Anthropic announcement, Stocktwits sentiment went to "extremely bullish" on "extremely high" message volume (Yahoo/Stocktwits). That was a hype spike, a negative under the rubric.
- Since then the "AI-cloud capex" framing has turned attention negative (Oct 8 downgrade, Oct 12 sector selloff). Wall Street Zen moved to Sell on Sept 26, then back to Hold on Oct 17. Retail attention is now moderate and fading. It is not viral, and not the low-attention ideal either.
4.5 Scoring build
| Component | Weight | Evidence | Sub-score |
|---|---|---|---|
| Mgmt-street divergence | 50% | Specific, repeated, quantified (≥$14B commitments, margin/FCF claims) vs flat consensus EPS, return-on-capex downgrades, 2x FY2028 EPS dispersion. Divergence is on economics, not demand | 8 |
| Mgmt track record (Dim 3 input) | context | Hit the top of the FY2025 revenue guide (low $4.0B, high $4.2B vs $4.21B actual); delivered the CIS 40% to 45% ARR target (Q4'25 CIS +45%); EPS beats every quarter but shrinking to $0.01 to $0.02 | supports |
| Analyst consensus | 20% | ~50/50 Buy/Hold, targets stale and far above price, active downgrades: mixed/skeptical | 7 |
| Insiders | 20% | CEO bought $6.6M in 2025 ($6.8M with the Chair); last 12 months $17.1M of open-market sales, no buys | 5 |
| Retail | 10% | Hype spike in Sept, fading now | 5 |
| Weighted | 0.5×8 + 0.2×7 + 0.2×5 + 0.1×5 = 6.9 | 7 |
What would move it: (+) Q3 RPO steps up by ~$11B+ and the CIS Q4 guide shows the hockey stick, plus a post-blackout CEO buy, gets to 8 to 9. (-) Q4 CIS ramp slips again, or Anthropic service start moves beyond H2 2027, and management's credibility on the economics breaks, giving 5.
Assessment
Akamai has a real, management-led divergence, not just a stock that is down. For four years Leighton and McGowan have repeated that distributed compute would be the growth engine, and they backed it with numbers: the 20% compute CAGR goal, the CIS 40% to 45% ARR target they hit, and now ~$14.2B to $14.4B of signed CIS contracts, ~$11.6B of them verified in the Sept 2026 Anthropic 8-K. At full ramp that is ~$2.1B/yr of contracted revenue against ~$314M of FY2025 CIS revenue. The street has accepted the revenue (FY2027 +16%, FY2028 +22% consensus) but not management's claim that these contracts earn low-20s to low-30s operating margins and strong post-build FCF. Consensus EPS is flat into FY2027, FY2028 EPS estimates range from $5.96 to $12.08, ratings are split ~50/50, and the downgrades (HSBC, the Oct 8 cut, the DA Davidson PT cut) all target capex returns and Anthropic concentration. That is the right kind of tension, and management's guidance record is good enough to give it weight. Two things hold the score at 7: insider confirmation is stale (the CEO's $6.6M of buying happened in 2025, and insiders have since sold $17.1M in the open market with no buying), and the Sept retail hype spike shows the story is not under-followed. The Nov 5 print, with Q3 RPO and the Q4 CIS hockey stick, will show whether management or the street is right about the economics.
Data sourced from Daloopa (company_id 251); market data and consensus from FMP (/stable, 2026-10-19); deal terms from SEC 8-K filed 2026-09-24; analyst actions and sentiment from cited web sources.