The Biggest Raise in Company History Is Real, But It Costs a Doubled Capex Plan and a Turn of Leverage

Results vs. Consensus

Equinix reported second-quarter results after the close on July 29 and held its call at 5:30 PM ET. The headline is a clean beat on both of the metrics the Street tracks, and a raise across the full-year guide (the ~51% margin and the dividend aside) plus a wholesale reset of the three-year outlook. The composition underneath is more interesting than the headline.

Q2 2026 Scorecard vs. Consensus

Q2 2026 Scorecard vs. Company Guidance

Consensus revenue landed within $1M of the company's own guidance midpoint. That is the recurring dynamic in this name and it is worth naming plainly: because Equinix guides quarterly revenue and adjusted EBITDA explicitly, and because the Street parks its estimate on the midpoint, the beat or miss headline is a report on where a lease-signing fee happened to land rather than on how the business performed. Last quarter that mechanism produced a headline miss on a quarter that was above the guide once the deferred fee was added back. This quarter it produced a beat.

Year-over-Year Comparison

Sequential Comparison

Revenue

Recurring revenue of $2,377M grew 10.9% as reported and 11% on a normalized constant-currency basis, the third consecutive quarter of double-digit recurring growth and a modest acceleration from Q1's 11.7% as-reported pace once the currency effect is removed. This is the line that matters, and it is doing what the thesis requires. Bookings support it: $424M of annualized gross bookings, up 23% year over year and the second-highest quarterly total the company has recorded, on top of roughly $110M of presales.

The composition of the beat is where the caution enters. Consensus expected $463.6M of interconnection revenue and got $453M. It expected $1.78B of colocation and got $1,772M. It expected $121.7M of managed infrastructure and got $112M. Every recurring line came in at or below the Street's mark, and the aggregate beat is entirely attributable to a fee that the company itself does not treat as recurring and that it now says should be modeled at roughly 5% of total revenue in steady state. At $2,625M of quarterly revenue, 5% is about $131M. Q2 carried $248M.

Margins

The reported adjusted EBITDA margin of 53.2% is a record and is up 314 basis points year over year, but roughly 230 basis points of that is the xScale fee flowing through at close to full incremental margin. The underlying expansion of about 150 basis points, on management's own framing, is the number to carry forward, and it is a genuinely good number: cash operating expenses grew 4.5% against 16.4% revenue growth, and general and administrative expense grew just 2.4%. Operating leverage is real and it is showing up in all three regions.

What the guide implies for the second half is the tension. First-half adjusted EBITDA of $2,641M on $5,069M of revenue is a 52.1% margin. The full-year guide midpoint of $5,240M on $10,245M is 51.1%, which requires second-half adjusted EBITDA of $2,599M on $5,176M of revenue, a 50.2% margin. The Q3 guide midpoint of $1,295M on $2,550M is 50.8%, which leaves an implied Q4 at roughly $1,304M on $2,626M, or 49.7%. That is not merely below the fee-inflated Q2 print. It is about 120 basis points below the ex-fee Q2 level of roughly 50.9%, on higher revenue, from a company that just told the market its operating leverage is compounding.

We flagged the identical arithmetic last quarter and it proved conservative: management guided a second-half margin below the Q1 level and then printed 53.2%. The difference this time is that there is a plausible mechanical cause. Equinix intends to double the cabinets it delivers in the second half and has pulled more than 7,000 cabinets out of 2027 into Q4 2026. New capacity carries operating cost before it carries revenue, and Q4, the quarter that absorbs the pulled-forward cabinets, is precisely the implied trough. Management gestured at this once, saying the expansion plan "minimizes earnings drag," and never quantified the drag or connected it to the guide.

EPS and AFFO

Diluted AFFO per share of $11.78 grew 18.9% as reported and 18% normalized. Below the operating line, the composition is deteriorating in a way that will matter more each quarter. Net interest expense was $115M against $83M a year ago, a 38.6% increase, as interest expense rose 11.9% on a larger debt stack while interest income fell 30.8% as cash was deployed into the build. That is a $32M year-over-year drag, roughly $0.32 per share, and management's own guidance that the blended cost of capital rises about 150 basis points by 2029 says the drag compounds from here.

Two items below the line went unexplained in both the release narrative and the call: $17M of impairment charges, against $2M in Q1 and $1M a year ago, and $28M of other expense against $1M of other income in Q1. Both are added back in the adjusted EBITDA and AFFO bridges. Neither is large enough to change the quarter. Both are the kind of thing a reader is entitled to have named.

Segment Performance

Revenue and Profitability by Region

Americas

Americas revenue of $1,251M grew 24.6% as reported, but $184M of it is non-recurring against $53M a year ago, and the Hampton lease sits here. Strip the roughly $120M of xScale fees and Americas revenue is about $1,131M, up 12.6%. Management said the same thing in its own words on the call, describing "low teens double digit growth" in the Americas after normalizing for Hampton. Recurring revenue of $1,067M grew 12.2%, the fastest of the three regions, and the adjusted EBITDA margin of 51.2% is up 482 basis points year over year, again fee-flattered.

Assessment: The underlying Americas number is the strongest in the portfolio on the metric that matters, and it is the region where the AI-driven demand is concentrated. It is also the region where the reported figures require the most work to interpret, because it holds the lease fees. Model the recurring line.

EMEA

EMEA revenue of $845M grew 10.2% as reported but only 8% normalized and constant currency, making it the slowest-growing region on a like-for-like basis despite the flattering headline. Recurring revenue of $806M grew 10.1% and, more importantly for our watch list, rose 2.2% sequentially from the $789M that had been exactly flat in Q1. That flat quarter was the single unexplained data point in the Q1 print, and it has resolved without comment. The adjusted EBITDA margin of 54.0% expanded 194 basis points.

"EMEA continues to form exceptionally well, not with that 2 of our main metros in EMEA, Frankfurt and Amsterdam are highly constrained metros."
— Adaire Fox-Martin, CEO and President

Assessment: The sequential resumption clears our open item from last quarter. But an 8% normalized growth rate is not "exceptionally well" relative to a portfolio compounding recurring revenue at 11%, and the constraint management names in Frankfurt and Amsterdam is a supply constraint, which caps volume even as it supports price. EMEA is the region where the sovereignty product set has to carry the growth, and Fabric Geo Zones is still in preview.

Asia-Pacific

Asia-Pacific revenue of $529M grew 9.1% as reported and 10% normalized, with recurring revenue up 9.6% to $504M. The region carries the highest adjusted EBITDA margin in the portfolio at 56.5%. It is also the only region whose adjusted EBITDA fell sequentially, to $299M from $305M, on revenue that rose from $526M to $529M, taking the margin down 148 basis points from Q1's 58.0%.

Assessment: A $6M, or 2.0%, sequential decline in absolute segment profit on rising revenue is worth an explanation, and none was offered. The CEO characterized the quarter as "an amazing quarter in APAC." On the recurring line APAC is the slowest of the three regions, and its margin moved the wrong way. This is the smallest region and one quarter is not a trend, but it goes on the watch list.

Revenue by Service Line

Key KPIs

Key Topics & Management Commentary

Overall Management Tone: This was the most forward-leaning call of the four quarters we have studied, and the confidence was structural rather than rhetorical: management framed the outlook raise as the output of a re-underwriting exercise, repeatedly volunteered the constraints on the plan, and put specific numbers on leverage, cost of capital, stabilization timing and metro concentration without being pressed. Where the call was thinner was on the second-order questions: the growth mechanism behind the revenue algorithm, the cash-conversion path, and what happens to the plan if demand disappoints all drew answers about capacity to accelerate rather than capacity to adjust. The call also ran short, eleven questions with no follow-ups permitted, on the day of the largest guidance raise the company has ever made.

1. The Largest Guidance Raise in Company History

Every line of the 2026 guide moved up, and management chose to characterize it in superlatives rather than let the table speak. Revenue guidance rose $100M before a $49M currency headwind, adjusted EBITDA $62M before $27M of currency, AFFO $50M before $18M. The revised ranges are $10,205M to $10,285M of revenue, $5,210M to $5,270M of adjusted EBITDA at roughly 51% margin, and $42.69 to $43.29 of AFFO per share.

"For 2026, we now expect revenue growth of 11% to 12% and AFFO per share growth of 10% to 12%. This is the largest single guidance raise in the history of our company."
— Adaire Fox-Martin, CEO and President

This is the second consecutive raise. The Q1 guide had already lifted the full year by $21M of revenue and $24M of adjusted EBITDA on Q1 outperformance. The pattern of a company beating its own guide and then raising is now three quarters old.

Assessment: The raise is real and it is broad. It is also worth holding against the margin arithmetic above: the full-year adjusted EBITDA margin guide did not move despite a 53.2% quarter, which means the raise is a revenue-and-dollars raise rather than a profitability raise. Management is banking the beat and holding the percentage.

2. The Long-Term Outlook Reset

The larger news was the 2027-2029 outlook, which replaced the framework given in June 2025 on every line. Annual revenue growth goes to 10-13% from 7-10%. The 2029 adjusted EBITDA margin target goes to 53%+ from 52%+. AFFO per share growth goes to 9-12% from 5-9%. Annual capital expenditures go to $5-7B from $3-4B. And dividend per share growth changes from "8%+" to "Approximates AFFO per Share Growth."

"Total revenue growth ranging from 10% to 13% per year beginning the period at the low end of this range and accelerating as the benefit of our capacity expansion plan builds."
— Olivier Leonetti, CFO

That sentence is the one that governs how much of the raise a 2027 model gets to keep. Growth starts at the bottom of the new band and accelerates into 2029, and the CFO added that AFFO should follow revenue. A 2027 AFFO per share growing at 9% to 10% off the 2026 midpoint lands at roughly $47, which is almost exactly where the top end of the previous 9-11% algorithm already put it. The three-year outlook improved materially. The next twelve months of it did not.

Assessment: This is a genuine upgrade to the terminal growth rate embedded in the multiple, and it deserves a re-rate on a long enough horizon. It is also, by management's own sequencing, a 2029 event financed by 2026 and 2027 capital. The dividend-growth line quietly lost its floor in the same table, replaced by a formula tied to a metric that itself begins the period at the low end of its range.

3. Capex Doubles, and the Return Disclosure That Came With It

FY26 total capital expenditures go to $5.0B-$6.0B from approximately $4.1B. The non-recurring component, which excludes xScale and real estate acquisitions, goes to $4,710M-$5,690M from a point estimate of approximately $3,800M. From 2027, the annual range is $5-7B.

"We continue to accelerate our capacity expansion to meet this growing demand In fact, we will double the number of cabinets we deliver in the second half of the year. As a result, we now expect 2026 CapEx to range between $5 billion to $6 billion."
— Adaire Fox-Martin, CEO and President

Last quarter we wrote that another material capex step-up without a matching return disclosure would be a downgrade trigger. Management pre-empted it, and did so more completely than we asked. The stabilized pool now returns 27% cash-on-cash on gross PP&E across 194 assets at 82% utilization, up from 26% on 192. The new capital was explicitly underwritten to the same standard, with the stabilization lag named.

"And we fully expect the new capital we are deploying to deliver the mid-20 percent yield you have grown accustomed to."
— Adaire Fox-Martin, CEO and President

"Another comment also, we think it is important in term of CapEx deployment at stabilization, which is about 3 to 4 years post-RFS, we expect to deliver the traditional 25% cash on cash returns that we have mentioned now for a number of quarters and years."
— Olivier Leonetti, CFO

Management also constrained the plan geographically, repeating the point three separate times: "More than 80% of this expansion will be our top 25 major global metros." The CFO added that a typical Equinix build "is about 60 megawatts," a deliberate contrast with the gigawatt-scale campuses elsewhere in the sector.

Assessment: The disclosure clears our trigger. It is the most complete return-and-scope framework Equinix has given, and the concentration into known metros with existing utility, contractor and community relationships is the right risk control for a plan of this size. What the disclosure does not do is change the cash arithmetic: a 3-to-4-year stabilization lag on $15-21B of cumulative spend means the returns being underwritten today first appear in the stabilized pool in 2029 and beyond, and the guide range for this year alone widened to $980M, which makes 2026 free cash flow unforecastable within about a billion dollars.

4. Cash Conversion, and What the Reported Free Cash Flow Line Hides

Reported first-half free cash flow was negative $560M against negative $607M a year ago, and adjusted free cash flow was negative $336M against negative $326M. Read alone, those lines say cash conversion was flat to slightly better while capex accelerated, which would be a remarkable outcome.

It is not what happened. Operating cash flow was $1,784M against $1,753M, up 1.8%. Purchases of other property, plant and equipment were $2,834M against $1,739M, up 63.0%. Operating cash flow less purchases of property, plant and equipment therefore went to negative $1,050M from positive $14M, a swing of $1,064M. The reported free cash flow line looks better because investing activities in 2026 carried $1,054M of short-term investment maturities against $789M of purchases, versus $450M against $795M a year ago, plus $348M of proceeds from asset sales that did not exist in the prior period. Portfolio liquidation and asset sales are not repeatable sources of funding.

The balance sheet shows where it went. Cash fell to $979M from $1,727M and short-term investments to $1,245M from $1,500M, so cash and investments dropped $1,003M in six months. Total debt principal rose to $22,153M from $21,417M. Net debt increased $1,739M. Net leverage nonetheless improved to 3.6x from 3.8x because adjusted EBITDA grew faster, which is the whole argument for the strategy working.

Assessment: Nothing here is alarming on its own. Net leverage fell, liquidity is $7.7B, the blended borrowing rate is 3.0% after tax and no equity was issued. But the underlying cash burn is now roughly a billion dollars per half year and it is set to widen, and the reported free cash flow line is a poor guide to it. Investors modeling this name off the free cash flow caption will underestimate the funding requirement.

5. The Interconnection Divergence

Equinix added a record 9,700 net interconnections, taking the base to 522,700, and interconnection revenue grew 11.3% year over year to $453M. Those two facts are harder to reconcile than they look. The base grew 1.9% sequentially and the revenue grew 1.6%. Revenue per interconnection, measured against the ending base, fell from about $869 per quarter to about $867. And for the first time in our coverage, interconnection grew more slowly than colocation.

The mechanism is visible in what management chose to talk about when the topic came up. Asked where the interconnection demand was coming from and whether there was room for pricing action, the answer moved immediately to the Fabric product suite.

"we can see some very significant growth rates in FCR, for example, our cloud router 170% up year on year in bookings there. And some of that driven through non colo customers, which is also an proposition for us."
— Adaire Fox-Martin, CEO and President

Virtual connections sold through Fabric and Cloud Router carry a lower price than a physical cross-connect. Growing units at 170% in a product priced below the legacy average is how you set a record for net adds while the revenue line decelerates. The CEO separately put interconnection growth at "around 9%" on a normalized constant-currency basis in the same breath as the record add number, and nobody returned to it.

Assessment: This is the single change in the quarter that touches the core of the thesis. The interconnection franchise is still the moat, and the ecosystem density argument is intact. But the specific mechanical claim we made at initiation, that interconnection outgrows colocation and therefore lifts mix and margin simultaneously, did not hold this quarter. If the mix shift is toward higher-volume, lower-price virtual connections, then the unit records and the revenue growth decouple permanently and the margin benefit is smaller than we underwrote. One quarter is not proof. It is enough to move the pillar to watch.

6. Bookings, Presales and the Backlog

Annualized gross bookings of $424M grew 23% year over year and 12.2% sequentially from Q1's $378M, the second-highest quarterly figure on record. Presales added roughly $110M, taking total sales activity above 30% year-over-year growth. Backlog set a second consecutive record.

"Starting with Serve Better, we delivered annualized gross bookings of $424 million up 23% year over year. A notable acceleration from Q1."
— Adaire Fox-Martin, CEO and President

The forward visibility disclosures were unusually specific: "we have already closed over 45% of our bookings target for Q3," and roughly 30% of the remaining 2026 retail capacity expansion has been pre-sold, up from about 25% at Q1.

The presale number itself moved the other way. Q1 delivered a record of approximately $140M; Q2 delivered approximately $110M. Management framed it exclusively on the year-over-year comparison, where it grew more than 50%.

Assessment: Bookings are the cleanest positive in the print and they lead recurring revenue by two to three quarters, which is what underwrites the second-half revenue guide. The presale sequential decline is a small thing in isolation, but presale visibility is a stated pillar of the 2027-2029 capex commitment, so it is the metric to watch rather than the one to explain away.

7. Churn and Pricing

Churn was 1.8%, a second consecutive quarter below the guided 2.0% to 2.5% band.

"Churn was 1.8%, primarily due to our renewal process execution and some delayed churn. We expect to be near the lower end of our typical 2 to 2.5 range for the back half of the year."
— Olivier Leonetti, CFO

Management continues to describe pricing as firm and, more interestingly, flagged an unrealized repricing opportunity in the constrained metros.

"we recognize that we are in a demand and supply continuum that is absolutely in our favor and we definitely see meaningful mark to market opportunity over the time period of our long range guide."
— Adaire Fox-Martin, CEO and President

Against that, MRR per cabinet of $2,538 grew 6% year over year on a normalized basis, decelerating from 7% in Q1.

Assessment: The churn commitment from last quarter is being met and we are grading it against the low end as we said we would. The mark-to-market comment is the most valuable unquantified item in the guide: if renewals in Frankfurt, Amsterdam and Ashburn reprice meaningfully over three years, the revenue algorithm has upside that is not in anyone's model. But MRR per cabinet decelerating while management describes pricing as firm and demand as exceeding supply is a mild contradiction that deserves a follow-up next quarter.

8. Financing the Plan

The CFO was the most specific he has been on how a $5-7B annual program gets funded, and the specificity was volunteered rather than extracted.

"We would expect to fund the growth through 2 levers: 1, retain cash flow. As you know, we have a payout ratio in the 50% range. So we will have a sizable retained cash flow and the rest of the financing will be done through debt."
— Olivier Leonetti, CFO

"we would expect leverage to increase by about a turn between now and the end of the planning period ... the blended cost of capital should increase by about 150 basis points."
— Olivier Leonetti, CFO

That is a debt-first hierarchy with no equity component, consistent with the commitment made at the CFO transition last quarter and consistent with the first-half cash flow statement, which shows $2,419M of senior notes proceeds and zero proceeds from any equity offering against $99M in the prior-year period. Investment-grade preservation was named as a core constraint: "Utilizing our balance sheet, we will achieve this with only a moderate leverage increase allowing us to maintain our current and critically important investment grade credit ratings."

The plan reconciles. Roughly $15-21B of cumulative capex over 2027-2029 against retained cash flow of roughly $2.2-2.9B per year at a 50% payout, plus a turn of leverage on a 2029 adjusted EBITDA in the $7-7.5B range, is enough to fund the top of the range without issuing shares. It does require the EBITDA growth to arrive on schedule, because the leverage capacity is a multiple of it.

Assessment: This is the most reassuring part of the call and it is why the stock rose. The reinvestment spread survives the change: 27% delivered against a 3.0% after-tax blended borrowing rate today, and even at the guided plus-150-basis-point cost of capital the spread stays north of 2,000 basis points. The dependency worth naming is circular. The debt capacity that funds the growth is sized off the EBITDA the growth produces, so a demand disappointment compresses the funding and the returns at the same time.

9. The xScale Fee, Finally Delivered and Finally Framed

The Hampton lease, which slipped from the Q4 2025 guide at $54M to the Q1 2026 guide at $80M and then out of Q1 entirely, signed in Q2.

"As expected, we closed 134 megawatts of [xScale] leases, including Hampton, which contributed approximately $120 million in nonrecurring fees."
— Olivier Leonetti, CFO

More useful than the fee itself is that management finally supplied the forecasting framework it declined to give when asked directly last quarter.

Q: "When we are looking at the new guide and kind of going forward, what is the right level to think of the normal for non recurring revenue and the long term guidance as a percentage of total revenue, how should we think about that with the new guidance level you set?"
— Frank Louthan, Raymond James

A: "You should assume the traditional 5% of total revenue that would be a good moderate assumption."
— Olivier Leonetti, CFO

Assessment: Commitment met and risk downgraded. A 5%-of-revenue steady-state assumption is crude, but it is a framework where there was none, and it lets a model normalize the lumpiness rather than guess at it. On that basis Q2's $248M of non-recurring revenue is roughly $117M of over-earning that will not repeat, which is another way of stating the ex-fee revenue growth of about 11%.

10. Supply-Side Risk Control

The most substantive answer of the call came on power, equipment and labor, the three constraints that decide whether a $5-7B program is deliverable at all.

"Today, we have 3 gigawatts of land under control. and we are building on about 700 on 700 megawatts of that land right now. We are not speculative land developers or land purchasers."
— Adaire Fox-Martin, CEO and President

Management described an internal gate before any project is announced, covering power contracting and energization plus permitting, which it credits for the on-time delivery record. On equipment, it said the balance sheet has been used to pre-purchase mechanical and electrical equipment, and that the design is fungible across the footprint so equipment can be redirected. On labor, it cited long-standing general-contractor relationships in North America.

Assessment: The gating discipline is the right answer and the pre-purchasing is a genuine use of the balance sheet as a competitive weapon. What was not given is what was asked: the percentage of incremental capacity backed by an executed utility energy services agreement. "A very high degree of confidence that power will be contracted" is a characterization, not a number, and power is the binding constraint on the entire plan.

11. The AI Demand Case, in Four Named Use Cases

Management sharpened the AI narrative from last quarter's aggregate statistics into a taxonomy: a stack use case where enterprises run open models on private infrastructure to cut token cost, a sovereign use case driven by data residency, a batch use case for training and batch inference enabled by liquid cooling, and a latency-sensitive use case where the inference stack must sit in-metro. Ecosystem density was quantified as "8 of the top 10 model providers as well as 8 of the top 10 neo clouds are already running their key networking workloads on Equinix today."

Product proof points followed: Fabric Geo Zones in preview with approximately 80 enterprises against roughly 20 last quarter, Secure Cabinet Express orders up more than 30% year over year, and expanded collaborations with Cisco and NVIDIA on standardized AI factory architectures.

Assessment: The taxonomy is a real improvement in disclosure quality and it makes the demand case falsifiable in a way that "approximately 60% of our largest deals in Q1 were AI-related" was not. Note, though, that this quarter's version of that statistic became "the vast majority of our largest deals were driven by AI workloads," which is a move from a number to a phrase. The direction is right; the precision went backwards.

Guidance & Outlook

FY 2026

Q3 2026

Long-Term Outlook, 2027-2029

Implied second-half ramp. The FY revenue midpoint of $10,245M against first-half actual of $5,069M requires $5,176M in the second half. The Q3 midpoint of $2,550M leaves $2,626M implied for Q4, a 3.0% sequential step. On adjusted EBITDA, the FY midpoint of $5,240M less first-half actual of $2,641M requires $2,599M, a 50.2% second-half margin, of which the Q3 guide takes $1,295M at 50.8% and Q4 is left at roughly $1,304M on $2,626M, or 49.7%. Q4 is the implied trough, and it is the quarter that absorbs the 7,000-plus cabinets pulled forward from 2027.

Implied second-half AFFO. First-half diluted AFFO per share of $22.57 against the FY midpoint of $42.99 leaves $20.42 for the second half, about $10.21 per quarter against Q2's $11.78. Recurring capex explains part of the step down: the FY midpoint of $300M against $81M spent in the first half leaves $219M for the second, roughly $110M per quarter against Q2's $49M, which is about $0.61 per share per quarter of incremental drag.

Street at: Consensus entering the print sat on the Q2 guidance midpoint and the full-year AFFO per share guide had been trending up. The revised FY26 AFFO per share midpoint of $42.99 is $0.28 above the prior midpoint of $42.71.

Guidance style: Conservative and consistently so. Equinix has now beaten the top end of its own quarterly revenue and adjusted EBITDA guidance and then raised the full year, twice in a row, while leaving the full-year margin percentage untouched. The pattern argues the implied 49.7% Q4 margin is a floor rather than a forecast. It also means the guide carries progressively less information about the actual trajectory.

Analyst Q&A Highlights

What changed in twelve months to justify the outlook reset

The call opened on the obvious question and it was really two questions: what in the market supports doubling the capex, and has the revenue algorithm shifted from pricing per cabinet toward cabinet count. Management answered the first at length and the second not at all.

Q: "I just wanted to get your view on the long term guidance raise Obviously, a huge change from last year. So maybe you could just talk through a high level of what you are seeing in the market that is given you the degree of confidence to raise CapEx this much And as we think about the forward growth mechanism for revenue of 10 to 13%, I think previously you had talked more about it being more based on MRR per cabinet growth than installed or billable cabinets. Has that changed at all based on the CapEx increase and based on the pipeline that you talked about?"
— Eric Luebchow, Wells Fargo

A: "So our revised outlook I think reflects more than a strong quarter It reflects a market opportunity that has materially improved over the past 12 months And this has been a huge collective effort"
— Adaire Fox-Martin, CEO and President

Assessment: The market-opportunity half of the answer was thorough and included the useful detail that inference has moved ahead of the internal schedule. The mechanism half never came back. Whether 10-13% revenue growth is priced or volume is not a modeling nicety, because it determines the margin path: pricing flows through at close to full incremental margin, cabinets do not. MRR per cabinet decelerated to 6% from 7% this quarter, which is at least suggestive of the answer management did not give.

Whether the capex plan can flex down as well as up

The most important risk question of the call asked directly whether the program can be adjusted if realized demand disappoints. The response described, in detail, the organization's demonstrated ability to accelerate.

Q: "Can you talk about the steps you are taking from an operational and risk management perspective to ensure that you can effectively deploy as much CapEx as you are budgeting over the next few years? And can adjust if realized demand does not match your forecast for some reason?"
— Nick Del Deo, MoffettNathanson

A: "In 2025, we were able to accelerate 20% of our retail footprint into 2026, And in our cabinet projections for Q4 of this year, you can see that we are almost tripling the number of cabinets that we will have available at that timeframe."
— Adaire Fox-Martin, CEO and President

Assessment: Acceleration capability is evidence of operational strength and it is not evidence of downside flexibility. The prepared remarks used the phrase "enables deployment flexibility in response to demand signals," which implies the ability to slow down, but the only worked examples offered in either direction were speed-ups. This is the question we would most like answered next quarter, because the answer determines whether the capex plan is a commitment or an option.

Financing the plan, leverage and the cost of capital

A question on renewal-spread contribution and financing tools produced the quarter's most quantitative disclosure, and it came from the CFO without prompting for numbers.

Q: "I am interested looking at the forward guidance in the contribution of things like renewal spreads. To the upside guidance on a multiyear basis. And then as we think about the CapEx plan going forward, what are the financing tools available to you? And how do you think about leverage"
— Jonathan Atkin, RBC Capital Markets

A: "we would expect leverage to increase by about a turn between now and the end of the planning period and to finalize the blended cost of capital should increase by about 150 basis points."
— Olivier Leonetti, CFO

Assessment: Both numbers are new and both are unfavourable, and management volunteered them rather than being cornered into them, which is the behaviour of a team that expects the plan to survive scrutiny. A turn of leverage on 2029 EBITDA is roughly $7B of incremental debt capacity, which combined with retained cash flow funds the top of the capex range without equity. The pricing half of the question, the mark-to-market opportunity in constrained metros, was answered qualitatively and remains the largest unquantified upside in the guide.

What the incremental capital buys, in megawatts

The most concrete attempt to convert dollars into capacity came late in the call, framed with the questioner's own arithmetic, and drew a one-sentence answer that ended the exchange.

Q: "I was wondering if you could talk about the 5 billion to 7 billion annual CapEx plan in terms of IT capacity I think over 3 years at $11 million per megawatt that would translate to about like 1.6 gigawatts out of the 3 gigawatts of developable capacity. Is that a reasonable way to think about it from an IT capacity perspective? And you know, how are you thinking about you know, refilling the land bank"
— Michael Ng, Goldman Sachs

A: "we will have about 2 gigawatts still available and this additional CapEx will use about 0.3 gigawatts of power."
— Olivier Leonetti, CFO

Assessment: The two figures in that answer are hard to hold together, and harder still against the 3 gigawatts of land under control cited earlier on the same call: 2 gigawatts remaining from 3 implies roughly a gigawatt consumed, not 0.3. The plainest reading is that 0.3 gigawatts refers only to the increment above the prior plan rather than the whole program, but that reading was not offered and no follow-up was permitted. The capital-per-megawatt translation is the single most useful check an investor can run on a plan of this size, and it remains unavailable.

Interconnection demand and whether there is room to price it

A question on interconnection asked where the demand was coming from and whether a product in this much demand supports incremental pricing action. The answer covered the product roadmap and did not address pricing.

Q: "if, in fact, you have a product that is in such high demand, how do you think about the opportunity for maybe incremental pricing actions on interconnect over the longer term?"
— Matt Niknam, Truist

A: "I think that there is the opportunity here to really look at the compelling value proposition that Equinix offers and then the opportunity to elevate that value proposition for our customers through some of these fabric offerings"
— Adaire Fox-Martin, CEO and President

Assessment: The offer to "elevate that value proposition" is a product answer to a price question. It is also, read alongside the revenue data, probably the honest one: the growth in the interconnection franchise is coming through Fabric and Cloud Router, which are priced below a physical cross-connect, so the pricing lever management actually has is upsell rather than list-price increase. That is a viable strategy. It is not the same strategy as the one that made interconnection the highest-margin, fastest-growing line in the portfolio.

Whether deal composition is changing as AI scales

A question on deal metrics, market demand and interconnect attach rates produced the quarter's clearest statement on deal density and the only spoken interconnection growth rate.

Q: "are you seeing any changes in underlying deal metrics or compositions Are different markets more in demand? What about deal sizes and interconnect attach rates, especially with the strong net adds this quarter?"
— Ari Klein, BMO Capital Markets

A: "the density of our deals is moving upwards as customers seek to secure the capacity that they need for their energy and their compute future. ... we had a very strong interconnection quarter adding over 9.7 thousand net adds and interconnection revenue growing at around 9%."
— Adaire Fox-Martin, CEO and President

Assessment: Rising deal density is a positive for revenue per square foot and a complication for the power plan, since density is what turns a capacity constraint into a power constraint. The more revealing part is the pairing of a record add number with a 9% growth rate in the same sentence, delivered as though the two were the same good news. They are not, and the question about attach rates went unanswered.

Supply-side gating on power, equipment and labor

The most comprehensive question of the call asked for the share of incremental capacity backed by an executed utility agreement, plus visibility on mechanical and electrical equipment and skilled labor.

Q: "how should we think about the percentage of the incremental capacity supported by this CapEx where there is an explicit ESA with the utility, and have visibility into that power. And then also if you could give us color into the visibility that you have on the M&E that you would need and as part of that the skilled labor to deliver the incremental capacity?"
— Michael Elias, TD Cowen

A: "all of those 3 gigawatts were either certain of our power and that it is contracted or that we have a high degree of confidence that power will be contracted, a very high degree of confidence that power will be contracted."
— Adaire Fox-Martin, CEO and President

Assessment: The equipment and labor halves were answered well, with pre-purchasing of mechanical and electrical equipment against the balance sheet and a fungible design that allows redeployment across sites. The power half was answered with an either/or that collapses contracted capacity and hoped-for capacity into a single reassuring category, and the repetition of "a very high degree of confidence" in place of a percentage is conspicuous in an otherwise numerate call.

The path to a 53% margin

Asked to lay out the puts and takes on multi-year margin expansion, the CFO gave a three-lever framework, which is the clearest statement of the operating algorithm the company has offered.

Q: "there any updates on how we should think about the puts and takes to the margin expansion over the next few years?"
— Cameron McVeigh, Morgan Stanley

A: "We are targeting a 53% plus EBITDA margin, which would be driven by 3 factors. 1 that Adaire has mentioned extensively which pricing. Demand is oversupply. So pricing would be a lever. And then cost of revenue improvement and SG and A scaling would be also 2 other levers."
— Olivier Leonetti, CFO

Assessment: Two of the three levers are already visible in this quarter's numbers, with cash operating expenses growing 4.5% against 16.4% revenue growth and general and administrative expense growing 2.4%. The functionalization and automation argument behind that is credible and it is the same argument that produced the underlying 150 basis points this quarter. The pricing lever is the one to test, because it is the lever that MRR per cabinet decelerating to 6% growth calls into question.

What They're NOT Saying

1. The second-half margin arithmetic, for the second consecutive quarter: the full-year guide implies a 50.2% second-half adjusted EBITDA margin and roughly 49.7% in Q4, below the ex-fee Q2 level of about 50.9%, on higher revenue. Management said the expansion plan "minimizes earnings drag" and never quantified the drag, never sized the pre-stabilization cost of doubling cabinet deliveries, and never reconciled the implied step down against the operating leverage it spent the call describing.
2. The real cash conversion: operating cash flow less purchases of property, plant and equipment went to negative $1,050M from positive $14M a year ago. The reported free cash flow line improved because $1,054M of short-term investments matured and $348M of assets were sold. Neither the release narrative nor the call mentioned the swing, and nobody asked.
3. Interconnection revenue against interconnection units: a record 9,700 net adds and a $10.6M shortfall against the Street on the same line, with growth crossing below colocation for the first time. The topic surfaced twice in Q&A and both times moved to the Fabric product roadmap rather than to the revenue line.
4. What the incremental capital buys: the only capacity translation offered was "about 2 gigawatts still available and this additional CapEx will use about 0.3 gigawatts," which does not reconcile with a 3-gigawatt land bank, and the call ended before anyone could ask again.
5. The dividend growth floor: the long-term outlook replaced "8%+" dividend per share growth with "Approximates AFFO per Share Growth." That change appears only in a table. It was not mentioned in prepared remarks and no analyst raised it.
6. atNorth: announced in Q1 as a definitive agreement with CPP Investments and described as immediately accretive to AFFO per share on close. No closing date, purchase price, contribution split or progress update in Q2. Two consecutive quarters unaddressed by management and unasked by the Street.
7. Presales fell from a record: approximately $110M against approximately $140M in Q1, in the quarter the company committed to a three-year capital program partly justified by presale visibility. Framed exclusively on the year-over-year comparison.
8. The share of capacity with contracted power: asked precisely, answered with "a very high degree of confidence that power will be contracted." Power availability is the binding constraint on the entire 2027-2029 plan and it is the one input given no number.
9. Whether the plan can slow down: asked directly, answered with three examples of the company's ability to speed up.
10. The below-the-line items: $17M of impairment charges against $1M a year ago, and $28M of other expense against $7M of other expense a year ago. Both are adjusted-EBITDA add-backs. Neither was named in the release narrative or on the call.
11. Asia-Pacific segment profit: adjusted EBITDA fell to $299M from $305M sequentially on rising revenue, a 148 basis point margin decline, in the region the CEO described as having "an amazing quarter."

Market Reaction

• Pre-print setup: EQIX closed at $1,008.02 on July 29, down 2.59% on the session ahead of the print. The stock entered the quarter up 31.6% year to date against the S&P 500's 6.9%, up 24.3% over the trailing twelve months, but down 7.1% over the trailing thirty days. The 52-week closing range was $726.09 to $1,115.94, so shares came into the print roughly 9.7% below the 52-week closing high.
• After-hours move: Shares rose about 2.2% to roughly $1,030 in the after-hours session following the release and ahead of the 5:30 PM ET call.
• Next-day session: The stock opened at $1,061.63, a 5.3% gap up, traded a $1,044.09 to $1,083.86 range and closed at $1,047.53, up 3.9% or $39.51. Volume was 0.9M shares against a 0.6M thirty-day average, roughly 1.4 times normal.
• Index: The S&P 500 rose 1.7% on the July 30 session, so EQIX outperformed the index by about 2.2 points.

The setup is the mirror image of last quarter's. In April the stock came into the print up 42% year to date and within 2.4% of its 52-week closing high, gave back the good news and closed the reaction session down 0.6% on a raise. This time it came in having given back 7.1% over the preceding month and roughly 10% off the high, and a comparable raise was rewarded with a 3.9% session. Almost nothing about the underlying business explains the difference in reception. Positioning does.

What is more interesting is the shape of the day. Shares opened up 5.3% and closed up 3.9%, giving back about a third of the gap. That fade is the market pricing the capex line. The initial reaction was to the raise and the record bookings, both of which are unambiguous positives available in the first two paragraphs of the release. The fade is what happens when the guidance table gets read and total capital expenditures for 2026 turn out to be $5.0-6.0B against approximately $4.1B a quarter ago, with $5-7B a year to follow for three more years.

Our read is that the return disclosure, not the raise, is what kept the day positive. A doubled capital program announced without the 27% stabilized yield, the reaffirmed mid-20s underwriting on new capital, the named stabilization lag and the explicit funding hierarchy would have been received very differently by a market that has spent two years re-rating data-center capital intensity. Management understood that and led with it.

Street Perspective

Debate: Is the capex doubling a compounding decision or a defensive one?

Bull view: The bull case being made on the Street is that a business earning 27% cash-on-cash on its stabilized pool should deploy every dollar it can source at a 3% after-tax cost, and that the constraint has always been capacity rather than demand. On this view the prior $3-4B plan was the anomaly and the raise simply lets the company stop turning away business in metros where demand exceeds supply.

Bear view: The bear camp contends that every operator in the sector is announcing a larger capital program at the same moment, that the 2029 payoff assumes today's demand curve holds through a full build cycle, and that the 3-to-4-year stabilization lag means the first honest read on these returns arrives after the capital is spent and irreversible.

Our take: The bulls have the better of this one on the evidence available, and the reason is the metro concentration rather than the yield claim. Putting more than 80% of the program into 25 metros where Equinix already operates, already holds the utility and contractor relationships, and already has an ecosystem to sell into is a materially different risk profile from greenfield gigawatt campuses. The 60-megawatt average asset size is the tell. This is a densification program dressed as an expansion program, and densification into an existing ecosystem is the highest-confidence capital Equinix can deploy.

Debate: Does the record interconnection add count still mean what it used to?

Bull view: A growing consensus view is that 9,700 net adds and 522,700 total interconnections extend a lead that no amount of capital reproduces, that Fabric and Cloud Router are widening the funnel to non-colocation customers, and that virtual connections land customers who later take physical capacity.

Bear view: Some sell-side desks argue that the unit metric has quietly stopped being a revenue metric, that a 170% growth rate in a low-priced virtual product is what produces record adds alongside decelerating revenue, and that the interconnection premium which justified Equinix's multiple relative to wholesale peers is compressing.

Our take: The bears are describing something real and the bulls are describing why it may not matter yet. Interconnection grew more slowly than colocation for the first time in our coverage, its share of recurring revenue fell sequentially, and it missed the Street. Those are facts, not framings. What is not yet established is whether virtual connections are cannibalizing cross-connects or funnelling toward them. One quarter of divergence is a data point. Three would be a re-rating.

Debate: Does the improved long-term algorithm justify the multiple?

Bull view: The bull case is arithmetic. AFFO per share growth moved to 9-12% from 5-9% and the 2029 margin target to 53%+, so at an unchanged multiple the stock is meaningfully cheaper against its own algorithm than it was a quarter ago, and a growth algorithm that improves by 400 basis points at both ends should attract multiple expansion rather than merely more earnings.

Bear view: The bear camp contends that the improvement is entirely back-loaded, on management's own sequencing that growth begins at the low end and accelerates, and that the near-term cost is explicit: a turn of leverage, 150 basis points of blended cost of capital, and a cash burn that widens before it narrows. A 2029 promise discounted at a higher cost of capital is worth less than the headline suggests.

Our take: This is the debate that sets the rating and we land closer to the bears on timing while agreeing with the bulls on direction. Our 2027 AFFO per share estimate is roughly $47, unchanged from last quarter, because the new 9-12% band begins where the old 9-11% band topped out. The algorithm genuinely improved. The next twelve months of it did not, and the next twelve months are what a twelve-month rating is about.

Model Update Needed

Valuation

Twelve-month framework. Applying a range of multiples to our FY27 AFFO per share estimate of approximately $47.00, against the July 30 close of $1,047.53:

The base case of $1,128 plus the 2.0% dividend yield implies roughly a 10% total return over twelve months. That is an improvement on the roughly 6% our Q1 framework produced, and it is achieved through a lower entry price and a better growth algorithm rather than through multiple expansion we are unwilling to underwrite. It is also, at approximately 10%, right at the return we would require of the index rather than clearly above it. The distribution has improved and remains close to balanced, with 16.7% of upside against 10.3% of downside.

Thesis Scorecard Post-Earnings

Overall: Thesis strengthened on funding and returns, weakened on the interconnection mechanism. The two pillars that moved are the two that matter most to how the multiple is justified. Bear #3 downgrading is a clean win on a commitment we set last quarter, and Bear #2 escalating is the price of the growth being underwritten. The net is a business whose long-term algorithm improved by 400 basis points and whose near-term cash conversion deteriorated by roughly a billion dollars, in the same print.

Action: Maintain Hold. The downgrade trigger we set at initiation, another material capex step-up without a matching return disclosure, was explicitly cleared and we credit management for pre-empting it rather than waiting to be asked. The upgrade trigger was not met: 24.4x is not the low-20s entry we said we needed, and the interconnection mix shift we wanted two more quarters of confirmation on went the other way this quarter. We would upgrade to Outperform on a de-rating toward 21x forward AFFO, or on two quarters of interconnection revenue re-accelerating above colocation. We would downgrade to Underperform if the 2027 capex lands above the $7B ceiling, if the blended cost of capital rises faster than the guided 150 basis points, or if interconnection revenue growth falls below 9% while unit adds continue setting records.