AKAM — Akamai Technologies, Inc.

The tape is the clearest signal: as of 2026-09-13 the stock was down 13.0% in a month and 20.0% in three, sitting below all three major moving averages with RSI at 43.57. The cause is the capital-intensity shock — management has guided 2026 capex to roughly 40-42% of revenue for GPU/edge capacity, non-GAAP operating margin is compressing toward ~26%, Q2 2026 free cash flow fell to $100.5M from $235.4M a year earlier, and GAAP EPS dropped 27% YoY. Sell-side positioning is genuinely split (25 Holds versus 26 Buy/Strong Buy; JPMorgan upgraded to Neutral with a $158 target in August, BofA/Oppenheimer stay constructive on underappreciated AI inferencing), and consensus targets of $137-$159 sit far above the tape. That gap between price action and price targets is itself the story: the market is refusing to capitalise the AI pivot until it shows up in revenue.

Revenue is stable, not deteriorating: YoY growth ran 4.1% (Q3 2024), 2.5% (Q4 2024), 2.9% (Q1 2025), 6.5% (Q2 2025), 5.0% (Q3 2025), 7.4% (Q4 2025), 5.8% (Q1 2026) and 5.4% (Q2 2026) — growth inflected upward through 2025 and has settled in the mid-5s, with Q2 2026 revenue of $1,099.7M a record. Margins are where the damage is, and it is recent and sharp: GAAP operating margin went 14.5% (Q4 2024) → 15.2% → 14.5% → 15.7% (Q3 2025) → 13.8% (Q4 2025) → 10.7% (Q1 2026) → 7.3% (Q2 2026), while gross margin slipped to 55.8% in Q2 2026 from 59.1% a year earlier (~330bps). R&D rose 18.3% YoY to $148.8M, deliberately. Free cash flow has declined four consecutive quarters: $246.8M → $161.9M → $120.7M → $100.5M. This is an inflection point, but a self-inflicted one — capital is being redirected into CIS/GPU capacity rather than lost to pricing or demand collapse. The verdict hinges on whether Security + CIS growth outruns the Delivery decline and whether capex intensity peaks; neither is yet visible in the numbers.

As of 2026-09-13 the stock traded at 37.6x trailing GAAP EPS of $2.841 and roughly 15.9x FY2027 consensus non-GAAP EPS of $6.716, versus its own 5-year average P/E of 28.89x — so the headline multiple looks cheap. That comparison is misleading: non-GAAP EPS is about 2.4x GAAP (TTM GAAP net income of $411M versus ~$977M implied by the non-GAAP number), with SBC and acquired-intangible amortisation making up the bridge. On cash, TTM free cash flow is $629.9M — a 4.1% yield — and TTM EV/EBITDA is 18.9x, which is not cheap for a company with flat guided EPS growth. P/S is 3.59x versus a 5-year average of 4.12x, P/B is 3.25x against D/E of 1.97 and LT debt/equity of 1.23. Net call: modestly cheap on forward non-GAAP earnings, fair-to-full on cash flow and EV/EBITDA — I would not call this the deep-value setup the 15.9x forward P/E implies.

The market is pricing flat earnings and an unproven capex bill at the same time. FY2026 non-GAAP EPS guidance of $6.40-$7.05 (midpoint $6.725) and FY2027 consensus of $6.716 imply essentially zero EPS growth, and at $106.79 the shares change hands at 15.9x that — well below where AKAM has historically traded. If the Q3/Q4 2026 prints show CIS revenue accelerating while capex intensity peaks, estimates have room to rise and the multiple can re-rate toward 20x or more, which is how the $137-$159 analyst targets get justified. If capex stays at 40-42% of revenue into 2027 without a revenue inflection, free cash flow stays near $100M a quarter, the balance sheet keeps levering, and 15.9x stops being cheap and starts being a value trap. Watch the CIS revenue line, the capex-to-revenue ratio and the GAAP-to-non-GAAP gap — not the EPS headline.

As of 2026-09-13, RSI(14) was 43.57 — neutral, neither oversold nor overbought. The $106.79 price sat 2.78% below the 20-day SMA, 7.47% below the 50-day and 2.90% below the 200-day, i.e. below all three. The stock is in the lower third of its 52-week range of $70.82-$165.45, down 13.0% over one month and 20.0% over three, though up 1.49% over the past week — a downtrend that is losing momentum rather than capitulating. There is no oversold bounce signal and no reclaim of the 50-day. Entry-timing implication: no urgency to buy; the first real evidence of a bottom would be a base forming above roughly $100 plus a reclaim of the 50-day moving average.

• AI-infrastructure capex race: AKAM has guided 2026 capex to ~40-42% of revenue for GPU/edge capacity, which makes the equity a levered play on GPU pricing, supply and power availability rather than a pure software-multiple story.
• Rates and cost of capital: long-term debt roughly doubled to $5.86B (from $2.40B at Q1 2025), lifting debt/equity to 1.97 and LT debt/equity to 1.23 — refinancing costs and any back-up in yields now matter to the equity story in a way they did not two years ago.
• Secular CDN decline: the legacy Delivery segment is in structural contraction as hyperscalers bundle their own CDNs, so AKAM must outgrow a shrinking base with Security and Cloud Infrastructure Services.
• Enterprise security budgets: zero-trust, API security and microsegmentation spending remains comparatively resilient, and is the main reason consolidated revenue growth is still positive in the mid-single digits.
• Sector rotation: capital-heavy 'AI infrastructure' names have de-rated relative to asset-light software as investors question the payback period on GPU capex — AKAM's -20.0% three-month move is a direct symptom.

• Q3 2026 earnings in early November 2026 — the key test of whether Cloud Infrastructure Services revenue is accelerating enough to offset Delivery declines while capex intensity stays on plan.
• Any guidance indicating the 40-42% capex-to-revenue ratio peaks in 2026 and steps down in 2027 — the single biggest lever on free cash flow and on the multiple.
• Commercial proof points on the new agentic-security and AI-intent frameworks and the expanded MuleSoft API-security integration, evidencing that the pivot is monetising.
• Further sell-side repositioning: JPMorgan's August upgrade from Underweight to Neutral with a $158 target shows how far price targets sit above the tape; more upgrades would be a sentiment trigger.
• Balance-sheet events — with $5.86B of long-term debt against $4.75B of equity, any refinancing, incremental debt raise or GPU asset-financing disclosure will be scrutinised closely.

• Capex without a revenue return: if GPU/edge capacity does not convert into CIS revenue, free cash flow stays depressed (four straight quarterly declines to $100.5M) while the debt taken on to build it remains on the balance sheet.
• Margin compression proves structural rather than transitional: gross margin has already slipped ~330bps YoY to 55.8% and GAAP operating margin has fallen to 7.3% from 14.5%; a permanent mix shift into lower-margin infrastructure could cap the recovery even if revenue grows.
• Legacy Delivery decay outrunning Security/CIS growth, pinning consolidated revenue growth in the low single digits and undermining the whole re-rating case.
• Competitive pressure from Cloudflare, Fastly and the hyperscalers' own edge, CDN and security stacks — in precisely the segments AKAM is pivoting toward.
• Earnings-quality and dilution risk: non-GAAP EPS of ~$6.72 rests against ~$2.84 of trailing GAAP EPS, with stock-based compensation and intangible amortisation covering the gap; the non-GAAP number flatters both margins and the forward P/E.

Not a buy at $106.79, but not a short either — this is a 'prove it' story. The security franchise is real, revenue is still compounding in the mid-5s, gross margin remains above 55% and the accruals ratio is negative at -1.64% (cash earnings exceed reported net income), so the deterministic 'Weak Fundamentals' band overstates the damage. Against that, GAAP EPS is down 27% YoY, free cash flow has fallen four quarters in a row, and long-term debt has more than doubled to $5.86B to fund a GPU build with no demonstrated payback. At ~15.9x flat FY2027 non-GAAP EPS the stock is not expensive, but with free cash flow yielding 4.1% and still falling, I want one or two quarters of evidence — CIS revenue acceleration plus a peaking capex ratio — before paying up. My own fair value is around $114, so risk/reward is mildly positive but not compelling enough to act ahead of the Q3 print.

Probability-weighted blend of two forward-multiple scenarios applied to FY2027 consensus non-GAAP EPS of $6.716: a 70%-weighted base case at 16x (=$107.5), reflecting flat guided EPS growth, a 55.8% gross margin, $5.86B of long-term debt and a 4.1% trailing FCF yield; and a 30%-weighted bull case at 22x (=$147.8) if CIS/AI-edge revenue inflects and capex intensity peaks in 2026. Cross-checked against a normalised-FCF lens (~$850M mid-cycle FCF at a 5.25% required yield ≈ $106-111), which is consistent with the blend and with my view that the deterministic band is too harsh rather than too kind.