Reading options flow in data center REITs

Why data center REITs generate strong options flow

Options flow in EQIX and DLR is structurally different from most REIT flow for three compounding reasons. First, the AI compute buildout has created a step-change in leasing demand that makes every hyperscaler capital expenditure announcement a potential catalyst for signed lease commitments, translating directly into AFFO guidance revisions. Second, unlike most real estate where supply is constrained by zoning or construction cost, data center supply is constrained by electricity: a campus with permitted buildings but no power interconnection approval is worthless, and those approvals now take years in primary markets. Third, the colocation hub model at EQIX generates recurring, high-margin interconnection revenue that compounds as each new tenant joining a facility makes the exchange more valuable to every existing tenant, a network effect with no analog in traditional real estate. These three dynamics create options flow patterns that look more like a semiconductor capital equipment name than a REIT: large call blocks on hyperscaler capex announcements, protective put spreads on power delay disclosures, and multi-session LEAPS accumulation during development pipeline updates.

The relevant financial metric for REIT options traders is adjusted funds from operations per share (AFFO). AFFO strips out non-cash depreciation charges, which are large in capital-intensive real estate, and adjusts for maintenance capex, giving a cleaner picture of the cash available for dividends and reinvestment than GAAP earnings. When EQIX or DLR reports AFFO per share ahead of consensus estimates and raises forward AFFO guidance, it is the equivalent of an EPS beat-and-raise in equities, and the call accumulation that follows has the same mechanics: institutions that were positioned via LEAPS calls before the print see their thesis validated and often add to the position in the weeks after the disclosure as the revenue recognition from signed-but-not-commenced leases begins flowing through AFFO.

Hyperscaler leasing demand: translating AWS, Azure, and GCP capex into signed leases

The most reliable single catalyst for call flow in data center REITs is a hyperscaler capacity announcement. When Amazon Web Services, Microsoft Azure, or Google Cloud discloses multi-billion-dollar data center investment plans for a specific geography, the data center REITs in that market receive the leasing inquiry flow within weeks. The translation from hyperscaler announcement to REIT AFFO impact runs through several stages, and experienced options traders position at each stage:

• Capex disclosure and the announcement trade: When a hyperscaler announces a capital expenditure program, Microsoft's multi-year commitment to European AI infrastructure, AWS's multi-billion investment in Northern Virginia expansion, call flow appears in EQIX and DLR immediately because the market prices the probability that a fraction of that capex lands in leased data center space rather than self-built facilities. Hyperscalers use a combination of owned campuses and third-party leased facilities; the mix fluctuates based on power availability, timeline requirements, and geographic constraints. When a hyperscaler's own data center development pipeline is constrained by permitting or power queues, third-party leasing accelerates
• Signed lease disclosure and the earnings beat: The confirmatory signal is a new lease signing disclosed in a REIT's quarterly earnings call. When EQIX or DLR reports that a named or unnamed hyperscaler has signed a new lease for a specified number of megawatts with a multi-year term and built-in annual escalators, the size of the signing relative to the development pipeline is the key metric. A hyperscaler lease covering 30–50% of a new development campus locks in AFFO for that asset before construction is complete, dramatically reducing the execution risk premium in the stock and creating a clean call accumulation environment in the following quarters as the lease commences and revenue begins recognition
• Pre-leasing rate as the forward demand signal: Data center REITs disclose the percentage of their development pipeline that is pre-leased, signed contracts on capacity not yet operational. When pre-leasing rates on the development pipeline are above 70–80%, it signals that demand is absorbing supply faster than it can be delivered, which supports pricing power on the remaining unleased capacity. LEAPS call accumulation tends to cluster around earnings calls where pre-leasing rates on the development pipeline are disclosed at levels above analyst expectations, because high pre-lease rates mechanically derisk the AFFO growth trajectory for the next four to six quarters of lease commencements

Power constraints: megawatts matter more than square footage

The defining constraint on data center supply in the mid-2020s is not land, zoning, or construction capacity, it is electricity. AI GPU clusters consume 10 to 100 times more power per square foot than traditional enterprise server rooms, and the grid infrastructure required to deliver that power at scale takes years to permit and build. Understanding power constraints is essential for reading data center REIT options flow because the constraint simultaneously limits new competition (protecting existing REITs' moat) and limits the REITs' own development pipeline execution (creating negative catalysts when power delays affect delivery timelines):

• MW capacity under development and powered capacity: The operational metric that institutions track is not square footage but megawatts, the actual electrical power capacity that can be delivered to tenants. Powered capacity under development, total powered capacity across the portfolio, and the MW pipeline planned for completion over the next 12–24 months are the metrics disclosed in REIT supplementals. When a REIT reports a step-change increase in its MW development pipeline, driven by new power interconnection approvals, successful grid upgrade agreements, or permits for new substations, call flow builds because MW capacity is the binding constraint and more of it directly expands AFFO growth optionality
• Grid interconnection queues and the competitive moat: In primary data center markets, Northern Virginia (NoVA), Silicon Valley, Chicago, Dallas, utility interconnection queues now extend two to four years. A new entrant attempting to build a competing data center campus in NoVA cannot obtain power interconnection approval for a timescale that is relevant to most tenants' lease requirements. This effectively freezes primary market supply at current levels except for operators who already have interconnection agreements in place. When news of extended interconnection queue delays is reported, either through utility filings or data center industry commentary, call flow in EQIX and DLR builds because the existing powered portfolio's value is directly enhanced by barriers to new supply entry. Conversely, when a large operator announces that it has successfully secured power interconnection for a new greenfield campus in a constrained market, it can trigger put flow in established operators because the competitive moat is partially eroded
• Power procurement strategy and nuclear optionality: Data center REITs and their hyperscaler tenants face long-term power cost and reliability questions that create specific options flow catalysts. The emergence of long-term power purchase agreements (PPAs) with renewable energy producers, wind, solar, and the more recent wave of data center PPAs with nuclear plant operators represents a structural shift in how power procurement works. When a REIT or hyperscaler announces a PPA with a nuclear facility, whether a restarting plant or a new small modular reactor (SMR) project, call flow appears across the data center space because nuclear provides 24/7 firm power without the intermittency risk of wind and solar, directly addressing the most serious operational constraint on AI cluster deployments. Diesel backup compliance risk, regulatory requirements to limit generator testing and operation at existing facilities, is a put catalyst for operators with aging backup infrastructure in markets tightening emissions standards

Colocation pricing: stabilized rates, hyperscale wholesale, and the AI premium

Data center pricing is not uniform, the market bifurcates into several distinct segments with very different flow dynamics. Understanding the pricing structure is essential for reading REIT earnings commentary and the flow that follows:

• Retail colocation pricing: Retail colo is sold to individual enterprises in small increments, single cabinets, partial kilowatts of power, with month-to-month or one- to three-year contracts. This is EQIX's core business in its legacy IBX (International Business Exchange) campuses. Retail colo pricing has historically been stable and escalating at 2–3% annually, but the AI demand wave has created significantly higher pricing power in markets where GPU cluster deployments are competing for the same physical space as traditional enterprise colo tenants. When EQIX reports retail colo pricing above the stabilized escalation rate, it signals that AI-adjacent demand is inflating the pricing environment for the entire colocation tier
• Hyperscale wholesale pricing and mark-to-market: Hyperscale wholesale leases are signed directly with major cloud providers for large blocks of power, typically above 10 MW per lease, with longer terms (10–20 years) and lower per-unit pricing than retail colo. The flow signal from hyperscale wholesale is the mark-to-market rent gap: the percentage difference between new lease pricing and the average rate on the existing portfolio. When new hyperscale leases are signed at 20–40% above the portfolio average rate, it creates a multi-year AFFO tailwind as existing leases expire and renew at higher rates, a compounding effect that institutions model as a revenue growth floor and express via LEAPS calls in the one- to three-year expiration window
• AI GPU cluster demand inflating pricing power: The arrival of dense AI GPU cluster deployments, which require 3 to 5 times more power per square foot than traditional enterprise computing and generate significant heat requiring specialized cooling infrastructure, has created a premium tier of colocation demand. Operators who have invested in liquid cooling infrastructure (direct-to-chip water cooling, rear-door heat exchangers) and high-density power distribution can command meaningfully higher rents than operators with traditional air-cooled facilities. When a REIT reports that a significant percentage of new leasing activity is in high-density power configurations above 20 kW per cabinet, versus the traditional 3–5 kW per cabinet, it signals that the AI cluster demand is being captured in the portfolio rather than flowing to build-to-suit competitors. This is a call catalyst because high-density leasing carries higher per-kW revenue and improves margin per square foot

Interconnection revenue: EQIX's durable moat

Equinix generates a revenue stream that most data center operators and all pure-hyperscale wholesale REITs cannot replicate: interconnection revenue. Interconnection fees are the monthly recurring charges enterprises pay to establish direct, private connections between their equipment and other tenants within an EQIX facility, cloud providers, network carriers, financial exchanges, CDN providers. Understanding interconnection is essential for understanding why EQIX commands a premium valuation and generates distinctive options flow patterns:

• Cross-connect revenue and the network effect: Each interconnection port (a physical or virtual cross-connect) generates monthly recurring revenue that is independent of power consumption. As the number of cloud on-ramps, network carriers, and internet exchanges present in an EQIX IBX facility grows, the value of being present in that facility increases for every existing and potential tenant, because the reach of their network expands with each new interconnection partner. This network effect creates a sticky, compounding revenue stream. When EQIX reports accelerating interconnection revenue growth, disclosed as a total interconnection port count and revenue per port, LEAPS call accumulation follows because the interconnection revenue has higher margins than colocation power and space revenue and compounds structurally as the ecosystem grows
• International Exchange (IX) revenue and geographic hedging: EQIX operates internet exchange points where carriers and content providers exchange traffic at peering locations. IX revenue is a smaller portion of total revenue but is an important signal of network ecosystem density. EQIX also generates a meaningful share of revenue in non-dollar currencies, Euro, British pound, Japanese yen, Singapore dollar, because its international IBX portfolio spans 260-plus locations across 70-plus countries. This creates a currency hedging overlay on EQIX options flow: when the U.S. dollar strengthens against the basket of international revenue currencies, EQIX AFFO guidance faces translation headwinds that can trigger put flow even when the underlying operational metrics are strong. Experienced flow traders separate the currency translation signal from the fundamental interconnection revenue signal to avoid misreading dollar-strength-driven put flow as a bearish thesis on data center fundamentals
• xScale and the capital-light model: EQIX's xScale joint ventures develop hyperscale data centers, large-format campuses serving major cloud providers, using third-party capital (sovereign wealth funds, pension investors) rather than EQIX's balance sheet. When xScale developments are completed and sold to the JV, EQIX receives development fee revenue and retains management contracts, then recycles the proceeds into higher-margin retail colocation development. This capital-light model for hyperscale allows EQIX to participate in the hyperscaler demand wave without the balance sheet strain of pure hyperscale development, while keeping more capital allocated to the higher-margin colocation and interconnection business. When xScale completions and JV capital recycling are disclosed in earnings, call flow validates the capital efficiency of this model

Geographic expansion: where constraints drive flow

Data center REIT options flow has a strong geographic dimension because supply constraints and demand dynamics vary dramatically across markets:

• Northern Virginia dominance and saturation: Northern Virginia (the Ashburn, VA corridor) is the largest data center market in the world by powered capacity, a position driven by the concentration of U.S. federal government cloud contracts, proximity to AWS's U.S.-East-1 region, and decades of carrier and exchange infrastructure investment. The market has reached a point where power interconnection constraints are severe, and new supply additions are measured in years rather than quarters. Existing powered EQIX and DLR capacity in NoVA is the most competitively insulated in either portfolio. When data center market reports show NoVA vacancy rates below 2–3% at the same time as hyperscaler demand surveys show strong U.S. East deployment requirements, it is a structural call catalyst for both names
• APAC: Singapore, Tokyo, and Osaka constraints: Singapore has imposed a moratorium on new data center construction, physical and power constraints on the island state mean that existing capacity has extraordinary scarcity value. EQIX's Singapore IBX portfolio is among the most valuable colocation real estate in the world on a per-MW basis. Tokyo and Osaka are experiencing similar dynamics: Japan's AI investment wave has driven demand that outpaces the ability of local utilities to extend grid power to new data center campuses. When news of Singapore moratorium extensions or Tokyo power constraint developments reaches the market, call flow builds in EQIX because its existing licensed capacity in these markets cannot be replicated on any near-term timeline. APAC pricing power is a recurring LEAPS call thesis that institution desks revisit every earnings season
• EMEA: Frankfurt, Amsterdam, and the power challenge: Frankfurt and Amsterdam, the two largest European data center markets, are facing structural power constraints driven by grid capacity limitations and renewable energy transition timelines. The Netherlands introduced moratoriums on new large data center developments in Amsterdam. Germany's power grid modernization is expanding capacity but on a multi-year schedule. For EQIX's European portfolio, these constraints have the same dual effect as in APAC: existing powered capacity becomes more valuable while development pipeline execution faces timing risk from power approval delays. DLR's Interxion European portfolio is similarly exposed to EMEA power constraints, a put catalyst when power delays extend development timelines and a call catalyst when new interconnection approvals are secured
• Secondary market expansion as the unconstrained opportunity: Markets like Phoenix, Dallas, Atlanta, Columbus, and Salt Lake City have attracted significant new data center investment because grid power is more accessible and land costs are lower. When DLR or EQIX announces greenfield development in power-unconstrained secondary markets, it signals that the development pipeline is being actively managed to redirect capacity additions to markets where execution risk is lower. Call flow appears because secondary market expansion extends the runway for AFFO growth beyond what primary market constraints alone would suggest

AI vs. crypto: how Bitcoin mining companies compete for the same infrastructure

One of the most structurally interesting dynamics in the data center and high-performance computing space is the competition between AI GPU cluster deployments and Bitcoin mining operations for the same scarce resource: high-density power in purpose-built facilities with industrial cooling infrastructure. This creates an options flow overlap that connects EQIX and DLR flow to Bitcoin mining companies transitioning to AI HPC hosting:

• Power-dense real estate as the shared asset: Bitcoin mining facilities are built around the same fundamental requirements as AI GPU clusters, large amounts of electricity delivered at high density, efficient cooling systems (whether air, immersion, or hybrid), and backup power infrastructure. A facility capable of running mining ASICs at 50 kW per cabinet is, with modest modifications, capable of hosting GPU clusters. When Bitcoin price declines make mining economics marginal, the converted hosting use case, renting power capacity and cooling to AI companies running training clusters or inference workloads, becomes more economically compelling. This optionality in Bitcoin mining facility assets is the thesis behind IREN's and CORZ's pivot to HPC hosting, and it drives the options flow interaction between crypto-adjacent names and data center REITs
• Iris Energy (IREN): power-first build model and the HPC transition: Iris Energy is a Bitcoin mining company that has built its business around securing low-cost, renewable-sourced power, primarily in Canada and the United States, and constructing purpose-built mining facilities optimized for high electrical density and cooling efficiency. The strategic pivot to AI cloud compute hosting (HPC) involves repurposing portions of the same physical infrastructure to run NVIDIA GPU clusters for AI inference and model training customers, generating revenue from the hosting contract rather than mining Bitcoin. The IREN options flow dynamic tracks two distinct narratives: the self-mining split (how much of its powered capacity is deployed in Bitcoin mining versus HPC hosting) and the power cost advantage (low-cost renewable electricity provides a structural cost edge for HPC hosting economics). When IREN discloses new HPC hosting contracts, particularly with named AI companies requiring GPU clusters, call accumulation reflects the market pricing the revenue predictability of hosting contracts versus the volatile Bitcoin mining revenue. Put flow appears when HPC contract execution is slower than guided or when Bitcoin price weakness reduces the option value of the self-mining allocation
• Core Scientific (CORZ): bankruptcy emergence and the NVIDIA-era pivot: Core Scientific emerged from bankruptcy proceedings and restructured its balance sheet with a focus on pivoting from pure Bitcoin mining to AI HPC hosting. The most significant announcement in CORZ's post-bankruptcy trajectory was a multi-year hosting contract with CoreWeave, a GPU cloud provider backed by NVIDIA investment, to provide power and cooling infrastructure for NVIDIA GPU clusters. This contract represented a structural pivot: CORZ's existing mining infrastructure, purpose-built for high-density power delivery and cooling, is being converted to host AI workloads at guaranteed contract rates, providing revenue predictability absent from mining economics. CORZ options flow is driven by the pace of this transition: when new HPC hosting contract announcements are made or existing contracts are expanded, call flow accumulates because each contracted megawatt represents years of predictable AFFO-like revenue. When the conversion pace disappoints, slower-than-expected facility upgrades, NVIDIA GPU allocation delays, or contract terms less favorable than expected, put flow builds as the market reprices the transition risk

Ticker-by-ticker flow frameworks

Each name in this group has a distinct flow structure driven by its specific business model, financial reporting metrics, and primary catalysts:

• EQIX (Equinix): The defining EQIX flow characteristic is the interconnection network effect as a valuation moat. Call accumulation in EQIX tends to be concentrated in LEAPS expirations, 12 to 24 months out, because the interconnection revenue compounding thesis requires multiple quarters to validate. The binary events that drive IV expansion in EQIX are primarily quarterly earnings calls (where interconnection port count growth, retail colo pricing mark-to-market, and xScale JV capital recycling are disclosed), management analyst days (where development pipeline capacity additions and new market entries are detailed), and hyperscaler capex announcements that imply new lease demand in EQIX markets. International revenue mix creates a dollar-index overlay: when DXY is rising, EQIX pre-earnings IV builds around currency translation uncertainty in addition to fundamental operational uncertainty, occasionally creating IV overcalibrated to macro currency rather than business risk. The most reliable EQIX LEAPS call setups occur when interconnection revenue growth is re-accelerating, retail colo pricing is above the stabilized escalation rate, and xScale proceeds are being recycled into colocation development rather than sitting on the balance sheet
• DLR (Digital Realty): Digital Realty's flow is more directly tied to hyperscaler capex cycles than EQIX's because its hyperscale wholesale business is the dominant revenue driver. The Interxion acquisition integrated a significant European colocation and interconnection portfolio, adding a dimension of EQIX-like network exchange assets to what had been primarily a hyperscale wholesale portfolio. DLR's data center fund joint ventures, partnerships with sovereign wealth funds and institutional investors to develop large hyperscale campuses, create disclosure events around fund launches, capital contributions, and completion announcements that drive call flow. When DLR signs a new fund JV with institutional capital, it signals that institutional investors find the data center development returns compelling, validates the hyperscale demand pipeline, and extends the development runway without balance sheet concentration. Put flow in DLR is more sensitive than EQIX to hyperscaler capex caution signals: when AWS, Azure, or Google Cloud management guides to more modest data center investment in a given geography, DLR's hyperscale wholesale occupancy in that market faces a more direct risk than EQIX's colocation business
• IREN (Iris Energy): IREN options flow is a hybrid of AI infrastructure positioning and Bitcoin mining volatility. The self-mining versus HPC hosting revenue split is the primary reporting metric, when the HPC hosting percentage of total revenue is growing, call flow reflects the market assigning a higher multiple to the predictable hosting revenue stream versus the volatile mining revenue. IREN's power-first build model, securing power contracts before constructing facilities, means the pipeline of future capacity is defined by signed power agreements rather than building permits, creating a distinct forward capacity disclosure dynamic. Call flow in IREN tends to appear ahead of contract announcements and cluster deployment confirmations, while put flow builds when Bitcoin price weakness threatens self-mining economics and HPC hosting conversion timelines are extended
• CORZ (Core Scientific): Core Scientific's post-bankruptcy emergence means its options flow is shaped by the transition narrative: the pace at which mining infrastructure is being converted to HPC hosting capacity and contracted with AI workload customers. CORZ has less diversification than IREN across markets and power sources, making it more concentrated in its HPC transition bet. The CoreWeave hosting contract is the anchor of the CORZ bull thesis, call flow reflects confidence in contract expansion and execution, while put flow reflects concern about contract concentration risk, facility upgrade costs, and the dependency on NVIDIA GPU allocation for HPC customers. CORZ is a higher-volatility, higher-beta expression of the HPC infrastructure transition theme relative to IREN, and its options carry correspondingly wider bid-ask spreads and more event-driven flow patterns around quarterly earnings and hosting contract disclosures
• GTLB (GitLab): GitLab is a DevSecOps platform, a source code management, CI/CD, and security testing suite, that is cloud-adjacent rather than cloud infrastructure. It is included in this cluster as coverage completeness for flow traders watching the hyperscaler and AI development platform ecosystem. GTLB's flow is driven by net revenue retention, the competitive positioning of its AI-assisted coding features (GitLab Duo) against GitHub Copilot, and the pace of enterprise contract expansions in the DevSecOps consolidation trend where enterprises are reducing the number of separate point-solution tools they maintain. Call flow in GTLB builds when enterprise deal ACV is growing from platform consolidation wins and AI feature attach is generating incremental contract value above the base DevSecOps subscription. Put flow appears when the GitHub/Microsoft competitive dynamic intensifies, particularly when GitHub Copilot's enterprise attach rate grows faster than GTLB's AI attach, or when enterprises with existing GitHub contracts choose to expand within Microsoft's ecosystem rather than consolidate to GitLab's unified platform

Reading call and put flow around specific catalysts

Understanding which events drive institutional positioning, and which side of the trade accumulates ahead of each catalyst, is the practical application of the frameworks above:

• Hyperscaler capacity expansion announcements: When a hyperscaler discloses a new multi-billion-dollar data center buildout for a specific geography in an earnings call or press release, the immediate flow in EQIX and DLR is typically call-dominated. The positioning appears in near-month-to-one-quarter expirations as traders price the probability of new lease disclosures in the next REIT earnings call, and in LEAPS expirations as longer-duration holders express the multi-year AFFO growth thesis. The call-to-put ratio in these flows is distinctive, often 3:1 or higher on the day of the hyperscaler announcement, and the strike clustering reveals whether the market is expressing a near-term lease-signing thesis (at-the-money to slightly out-of-the-money calls in near expirations) or a long-duration AFFO growth thesis (deeper out-of-the-money LEAPS where the expected move is driven by a multi-year compounding scenario)
• Power constraint disclosures and put flow: When a REIT's management team discloses in earnings or a quarterly supplemental that a development campus has encountered power interconnection delays, whether from utility capacity constraints, grid upgrade timelines, or permitting challenges, put flow appears in the one- to three-month expiration window. The structure is typically a put spread rather than outright puts because the downside is bounded by the value of the existing powered portfolio, which is not impaired by a single development delay. When power delays affect multiple development projects simultaneously, the scenario that occurred in NoVA markets during periods of peak utility interconnection queue congestion, the put flow is more aggressive because it implies AFFO guidance risks are distributed across the pipeline rather than isolated to a single asset
• Development cost overruns and construction inflation: Data center construction costs are sensitive to the cost of transformers, switchgear, liquid cooling equipment, and skilled labor, all of which experienced significant inflation in the post-pandemic period. When a REIT discloses that development cost per MW has increased above prior guidance, driven by supply chain constraints on critical electrical equipment or labor cost escalation, put flow appears because higher per-MW cost reduces the return on invested capital for the development pipeline, compressing future AFFO accretion per development dollar. The most sensitive disclosure is when a management team revises per-MW development cost guidance upward while simultaneously holding AFFO guidance flat, the implicit signal that the pipeline returns have been diluted, which drives put accumulation in the two- to four-quarter expiration window as the market prices a multi-quarter AFFO headwind from development economics
• HPC hosting contract announcements in IREN and CORZ: For the Bitcoin mining-to-HPC transition names, the cleanest call catalyst is a new multi-year HPC hosting contract announcement with a named AI company or GPU cloud provider. The structure of the call flow reflects the contract's financial terms: when a hosting agreement is disclosed with high MW committed, multi-year term, and guaranteed minimum revenue, LEAPS calls accumulate because the contract converts volatile mining revenue into predictable infrastructure revenue that the market values at a higher multiple. The strike placement in these calls reveals whether institutions are expressing the near-term revenue certainty (at-the-money calls for the next four quarters) or the long-duration HPC infrastructure thesis (out-of-the-money LEAPS expressing the scenario where the entire portfolio completes its HPC transition)

Summary

Data center REIT options flow is governed by a distinct set of physical and financial constraints that separate it from traditional real estate sector flow. AFFO per share is the primary financial metric, the REIT equivalent of an EPS beat, and the catalysts that move it are hyperscaler lease signings, powered MW capacity additions, interconnection port count growth, and development pipeline pre-leasing rates. Power constraints in primary markets create a structural competitive moat for existing capacity holders that manifests as call accumulation on interconnection queue delay news. EQIX's colocation hub and interconnection network effect model generates the most durable LEAPS call setups because the compounding is mathematically predictable once the network density of a facility reaches a critical mass. DLR's hyperscale wholesale focus makes it more sensitive to hyperscaler capex cycle signals, the clearest read-through from Azure and AWS quarterly disclosures. IREN and CORZ represent a higher-beta expression of the AI infrastructure demand theme through the lens of converted mining infrastructure, where the HPC hosting transition pace is the binary catalyst. The geographic dimension, NoVA supply constraints, Singapore moratoriums, Tokyo power limitations, Frankfurt and Amsterdam development restrictions, adds a market-by-market layer to sector-wide flow that explains why the same AFFO growth trend in EQIX can produce different flow intensities depending on whether the geographic concentration of new development is in supply-constrained primary markets or power-available secondary markets. Reading this group's options flow well requires integrating physical infrastructure constraints, hyperscaler capex disclosures, and REIT-specific financial metrics into a single framework, the traders who do that successfully consistently position ahead of the major call and put accumulation events in the sector.