In both businesses the rivals are either faster or bigger.
Is it a good business? · 6 · What does its record show?
Slow growth, and profit per $1 nearly halved
Sales, $B
3.5
3.6
3.8
4.0
4.2
4.4
Dec 2021Dec 2022Dec 2023Dec 2024Dec 2025last 12 mo
GrowthAbout 4%revenue growth a year since 2022
Profit18.7% → 10.5%operating profit per $1
Share pay11.8%of revenue paid in shares, doubled
The cost of servers and space for the computing business rose ahead of its revenue.
Is it a good business? · 7 · How much room is there to grow?
Cloud renting is huge; Akamai's slice is unmeasured
$143Bspent renting cloud computing, second quarter of 2026
43%growth in a year
165%growth in the AI part
Nobody publishes how much runs in many small sites, the kind Akamai builds.
Is it a good business? · 8 · Who is in charge?
A co-founder on a $1 salary
$1Tom Leighton's salary as chief executive
About 1.9%of the shares held by Leighton
Dan Hessechairs the board
Edward McGowan is finance chief. Leighton co-founded Akamai and is paid almost entirely in shares.
The story right now · 9 · Why is Akamai renting AI computing?
AI firms need computing near their users
More than $14Bsigned contracts by October
+34%rented computing last quarter
Anthropicamong the customers
AI firms rent huge amounts of computing. Akamai is using its spread-out network to win some, signing long contracts in a few months.
The story right now · 10 · Do the AI contracts earn their cost?
Servers come first, payment later
A contract is signedOver $14B so far.
Akamai buys serversAbout $5.5B.
It borrows to payLoans that can turn into shares.
Rent comes inAt returns barely above their cost.
Management calls the returns healthy, but gives no profit figure for the computing business.
The story right now · 11 · What changed lately?
Rented computing grew 34%, slower than before
$1.12Brevenue, up 6%
$109Mrented computing, up 34% from 39%
About $30Mof fourth-quarter sales slipped into 2027
10% vs 25%official profit against adjusted
Mixed quarter
Security held. The jump on the first AI contract has gone, and October's fall below the pre-deal level is unexplained.
The story right now · 12 · What is everyone arguing about?
Will the AI contracts earn what they cost?
Many investors think
AI contracts make it a faster grower
We think
The contracts may barely earn their cost
Against us: customers must pay whether or not they use the computing, and management calls the returns healthy.
The price and our view · 13 · What is a share worth?
Counting the debt, the price is well above our value
$54.15what we think a share is worth · the last close is of that
$54.15our value
Attractive below $43.32
Neutral
Unattractive above $64.98
$7.84gloomy: growth halves and less kept per $1
$54.15our estimate: experts' path, minus all debt
$94.46sunny: growth beats forecasts by a quarter
Value per share in three futures. None is a prediction of the share price.
We estimate what a share is worth from the cash the business should earn over many years. Our rule: buy only when the price is at least a fifth below our estimate, as a cushion for being wrong.
The price and our view · 14 · How could we be wrong?
One customer, and a cash plan that runs short
One customer dominatesBiggest
It can end a plan after an outage.
Cash runs shortShowing now
2027 needs new money.
Loans turning into sharesWatch
$7.6B of them.
What computing earnsUnknown
No profit figure for it.
Delivery keeps shrinking, and bigger clouds compete for the AI work.
The price and our view · 15 · How will we know?
February tests rented computing and the funding
Rented computing reaching $120M a quarterGood sign
A clear plan to pay for 2027Good sign
Selling new shares to fund serversBad sign
FEB 202718
Fourth-quarter 2026 results, 18 February. Companies report every three months; each report is a test of the story.
Our view also changes on price alone: if the shares fall to within a fifth of our estimate, we move to Neutral.
The price and our view · 16 · So, what do we think?
A steady security network staking its balance sheet on AI
$54.15our value
Attractive below $43.32
Neutral
Unattractive above $64.98
UnattractiveMedium conviction
Akamai's security business is steady and profitable. But renting computing to AI firms needs heavy spending first, and $7.6B of loans that can turn into shares counts against owners. The price sits well above our value.
↑ ↓ to move · C for contents
What it does
More about the company
Who it is
Akamai runs servers in about 700 cities that sit between a big company's website or app and its users. Each visitor connects to a nearby Akamai server, which blocks attacks and bad bots, keeps copies of pages and video nearby, and passes on the rest.
It also rents out computing, ordinary processors and AI chips, in those places. In 2026 that grew into large contracts: Anthropic, the AI company, will rent servers Akamai is buying for it. The alternative is your own network, or Amazon, Microsoft or Google.
How it makes money
Customers sign contracts, usually for several years, and pay for the security, traffic and computing they use. In Q2 2026 security brought in $604.4M, 55% of revenue; delivery and older cloud tools $395.9M; and rented computing $99.3M, the fastest growing at 39%. No customer is a tenth of revenue. The large AI contracts work differently: the customer commits to pay for capacity for seven years once Akamai delivers it.
Growth and profit
What growth costs
Growth has cost margin. The official operating margin fell from 18.7% in 2022 to 10.5% over the last year, as pay in shares doubled to 11.8% of revenue and the costs of space, power and servers for rented computing arrived before its revenue.
Its record
The numbers
Is it growing?Slow
Sales grew 6.2% over the twelve months to Sep 2026, about the same pace as the 5.4% in the year to Dec 2025.
Is it profitable?Thin
Operating margin is 9.1%, down from 17% in the year to Dec 2023.
Does growth pay off?Costly
Running costs take 47¢ of each sales dollar, up from 41¢ in the year to Dec 2021.
Does it earn good returns?Low
Shareholders' money earned 7.9% a year, down from 15% in the year to Dec 2021.
Is it making cash?Solid
It turned 12% of sales into free cash, $545M over the last twelve months, after spending 20% of sales on buildings and equipment.
Is the balance sheet safe?Heavy debt
It owes $4.05B more than the cash it holds, about 2.8 years of the cash its operations bring in.
Are shares being diluted?Shrinking
The share count fell 8.8% in 4.8 years, and pay in shares costs 12% of sales.
Why it wins
The evidence
Akamai's edge is size. It has servers in more than 130 countries, joined to about 1,200 other networks, so it can carry and check a huge site's traffic close to its users, and seeing that much traffic sharpens its view of attacks. A US data breach costs about $10.22M on average, so big buyers value a shield that works.
It is not the most common shield: an outside survey finds Akamai on 1.9% of websites whose shield it can identify. Its customers are fewer and much bigger.
Who could beat it
The rivals in full
Cloudflare sells the same security and delivery from one network, grows far faster and sits in front of most identifiable websites. Fastly competes in delivery and names Akamai as a rival. In renting computing, Amazon, Google and Microsoft are far larger, and Akamai's own annual report warns they get first call on servers, memory, space and power. Anthropic still buys most of its computing from Amazon.
Large blocks of AI computing rented to model companies on multi-year commitments.
How big it could get
The market and who has signed
The market
Companies spent $143B renting cloud computing in Q2 2026, 43% more than a year before, and the AI part grew 165%. Nobody publishes how much of that runs in many small sites, the kind Akamai builds, so its own pond cannot be measured exactly. Against the whole market, Akamai's contracts, about $2.2B a year once running, are well under 1%.
Who has signed
Four companies signed this year: about $200M over four years, $1.8B over seven, more than $600M over four, and Anthropic's $11.6B over seven, about $1.7B a year once running. Anthropic must pay once Akamai delivers, but may end each plan after a material outage.
Who runs it
Leaders, pay and the cash
Who runs it
Tom Leighton, a co-founder, is chief executive and takes a salary of $1; he holds about 1.9% of the shares, and all directors and officers about 2.3%. Edward McGowan is finance chief and Dan Hesse chairs the board. Bonuses are paid in shares and reward revenue and adjusted profit; nothing rewards cash flow or the return on money spent, which matters now the company is spending billions on servers. Insiders sold over the year and bought nothing.
What they did with the money
Akamai spent about $616M buying back shares in the first half of 2026, mostly at prices well above today's, then stopped; it bought none in the third quarter. In May it borrowed $3.5B through bonds that can turn into shares, to pay for servers, and in July it bought LayerX, a browser security company, for about $205M.
The last report
The latest report
Akamai's third-quarter revenue grew 6% as rented computing grew 34%
What it did to the thesis Mixed quarter
It showed security holding and rented computing growing fast but slower than the pace we tested for, with about $30M of Q4 sales slipping into 2027. The call said cash will fund most of it. The new targets lift expectations; whether the contracts earn their cost stays open.
The report
In Q3 2026 revenue grew 6% to $1.12B. Security grew 9% to $622M, delivery and older tools fell 3%, and rented computing rose 34% to $109M, slower than the 39% before. Adjusted profit per share fell 5% to $1.77.
The cost of growth showed again: official operating profit was 10% of revenue against 25% adjusted, and Akamai plans $2.45B to $2.65B of server and memory spending in Q4 for its largest contract. It now guides 2027 revenue growth of 17% to 22% and 2028 revenue above $7.0B, and cut the top of its 2026 range.
What's happening now
The background
The shares jumped by about a quarter in one day in May on the first large AI contract, and had fallen below where they stood before it by October, even though the signed contracts grew to more than $14B. No filing or release explained the October fall.
The next test is the December-quarter report in February 2027: whether rented computing reaches $120M, and how 2027 will be paid for.
Where we disagree
Both sides in full
The market
The AI contracts turn Akamai into a faster-growing company, worth more than a slow security and delivery business.
We believe
The contracts will add revenue, but at about a third of a dollar a year per dollar of servers they may barely earn their cost.
The argument
The market counts the contracted revenue. We count what it costs: about $5.5B of servers before the big contract pays, a cash plan that runs short in 2027 without new money, and one customer who can end a plan after a material outage.
The evidence against us is real: the contracts are take-or-pay, management says their margins are healthy, and the shares already fell below where they stood before the first one.
What it's worth
How we value it
Valued on the analysts' revenue path for three years, each later year closing 30% of the gap to 4%, on official operating profit with stock pay a cost and tax paid in full; the thesis's own evidence is the check on that path.
What the price assumes
Sales growth in the first year, then slowing as ours does, at the same profit per sale.
The price assumed
16.5%
Our estimate
14%
Measured at the close of
6 Nov 2026
The three cases
Bear case$7.84
Growth runs at half the analysts' path, 7%, 11% and 12%, the margin settles a fifth lower at 9.9%, and new capital earns only the 8% rate: the downturn test.
Our value$54.15
Revenue grows 14%, 22% and 24% over three years, the analysts' path (C49), then fades toward 4%, closing 30% of the gap each year; the margin moves to 12.4%, the analysts' year-three figure after stock pay and on the filed basis (2.8% less).
Bull case$94.46
Growth runs a quarter above the analysts' path, 18%, 27% and 30%, the margin reaches 13.7%, and new capital earns 15%.
The bet
Seen as a bet
The bet is on Akamai's new business of renting computing to AI companies. The contracts are signed, mostly with Anthropic, but they pay only once Akamai has bought and installed the servers, most of them in 2027.
Valuing the rest of Akamai at a fair price leaves a large slice of today's price for this business. Even if every contract runs as planned, it is worth less than that slice: an expensive bet. If it fails, shareholders keep the established network, worth most but not all of today's price.
What could go wrong
Every risk
The risks
One customer dominates
Anthropic will be most of the computing contracts, and may end a plan after a material outage or an uncured breach.
Cash runs short
On the company's own plan, cash and the credit line do not cover 2027's server spending and bond repayment without new money.
New shares from bonds
Bonds due 2033 turn into shares near today's price, and Anthropic's warrant covers about 7.7 million shares.
Delivery keeps shrinking
Delivery revenue fell 6% in Q2 2026 on price pressure and customers building their own networks.
Bigger clouds compete
Amazon, Google and Microsoft get first call on servers, memory, space and power, Akamai's own annual report warns.
Pay in shares
Pay in shares, about $510M a year, roughly equals the cash left after all spending on servers and software, and buybacks have stopped.
What we still do not know
What rented computing earns after space, power and depreciation; Akamai reports its revenue but not its profit.
Whether the 2027 build-out stays funded by debt: $2.2B of convertible notes priced on 9 November, and shares are issued only if the price passes $102.
Whether the 2027 and 2028 revenue targets, which management gave on 5 November, are met on the contracts' schedule.
What changes our view
Every route to a new view
Every route to a new view
→ Neutral
The share price falls to about $64.98, back within 20% of what we think the business is worth.
→ Attractive
The share price falls to about $43.32, 80% of what we think the business is worth.
The next test
It supports the case if
Cloud infrastructure revenue of $120M or more, security growth of 8% or more, and 2027 funding named as debt or customer prepayments.
It pressures the case if
Cloud infrastructure revenue below $120M, security growth below 8%, or 2027 funding that includes new shares or is not named.
We would also be wrong if
Security revenue grows less than 5% from a year earlier for two quarters running.
Revenue from the Anthropic contract has not started by the end of 2027.
Anthropic ends a plan, or Akamai discloses a material outage.
The verdict
Scores and how we weighed it
The scores
Growth
2 of 4
Business quality
3 of 4
Financial position
2 of 4
Execution record
3 of 4
Per share capture
2 of 4
Potential
3 of 4
Risk profile
2 of 4
Valuation
2 of 4
Evidence quality
3 of 4
How we weighed it
ValueNegative
The price is well above our value once the convertible debt is counted, and near what the business is worth if the AI contracts all run as planned.
BusinessMixed
Security is steady and profitable and customers stay; delivery is shrinking, and the balance sheet is stretched by the build.
GrowthMixed
Signed contracts support faster growth from next year, but the price already needs growth well above anything the company has delivered.
RisksNegative
One customer is most of the new contracts and may leave after an outage, cash and credit do not cover next year's servers and bond repayment, and the convertible bonds add shares.
RivalsMixed
It is cheaper than its two filed rivals, but both grow several times faster, so the gap reflects slower growth, not a bargain.
The groupMixed
Security names move together on spending news; what sets this one apart is the computing build, which no other member of the group carries.
Latest newsMixed
The last report showed security holding and rented computing growing, a little slower than we tested for, and new bonds confirmed the build is paid for with debt.
In plain words
Our estimate of value
We forecast how much cash the business can earn over the next ten years and beyond, then ask what that stream is worth today to an investor who wants an 8% yearly return. We do it three times: a gloomy future, an expected one and a sunny one. We call a share Attractive only when its price is at least 20% below the expected estimate, and Unattractive when it is 20% above.