PayPal: The Dot-Com Crash IPO Anomaly
The Recursive Density Threshold
In February 2002, amid the radioactive fallout of the dot-com crash, a company processing payments for Beanie Babies and vintage electronics completed an initial public offering that valued the firm at $1.2 billion. The timing seemed almost perverse. Consumer internet companies had become toxic assets; the NASDAQ had shed nearly eighty percent of its value over eighteen months. Yet here was PayPal, pricing its shares at thirteen dollars each and watching them climb twenty-five percent on the first day of trading. The market saw something that accounting statements obscured: a recursive network density that, once achieved, created economic gravity difficult to reverse, provided the company could survive long enough to amortize the massive fixed costs of its infrastructure across sufficient transaction volume.
PayPal did not function as a traditional payment processor, nor as a bank, nor as a simple software layer. It operated as a two-sided recursive network where every node served dual purposes. A user who signed on to pay for a Pez dispenser on eBay simultaneously became a potential recipient for future transactions: merchant waiting to happen. Unlike credit card networks that connected distinct populations of merchants and consumers through an intermediary banking layer, PayPal’s architecture collapsed the distinction. Each new account holder increased the platform’s utility exponentially rather than linearly, because they brought with them both demand for payments and capacity to receive them. This density created the conditions for viral growth, but it also imposed a brutal economic prerequisite. The infrastructure required to maintain this network, a href=”https://bangid.com/product/” target=”_blank” rel=”noopener”>fraud detection systems capable of processing millions of daily transactions, data centers hardened against attacks, regulatory compliance architecture spanning multiple jurisdictions, epresented fixed costs of staggering scale. Unlike software companies that could serve ten or ten million users with marginal engineering adjustments, PayPal’s infrastructure costs stepped up in discrete, expensive increments.
The transaction economics offered terrifyingly little room for error. PayPal collected a take rate measured in single-digit percentage points on each transfer. After interchange fees paid to underlying banking networks, fraud losses, customer acquisition costs, and the operational overhead of maintaining the technical infrastructure, the margins were tissue-thin. Only at massive volume, billions of dollars flowing through the system annually, did the unit economics turn positive. The business was essentially a high-fixed-cost utility disguised as a nimble internet startup. Without the density threshold, millions of active users conducting diverse transactions, the fixed costs would consume the company. This structural reality shaped every strategic decision that followed, from the desperate need for viral distribution mechanics to the dangerous dependency on a single source of transaction flow, and ultimately to the cultural conflicts that would nearly destroy the company’s innovative capacity.
Captive to the Auction Block
The density threshold might have remained theoretical without eBay. At the time of the IPO, approximately sixty-six percent of PayPal’s payment volume flowed through the online marketplace. eBay had not built its own payments infrastructure; sellers and buyers had gravitated toward PayPal’s “send money” function as the least-worst alternative to mailing checks or money orders. In fact, sixty percent of eBay transactions still settled through analog methods, aper checks sent through postal mail, cash concealed in cardboard boxes, postal money orders purchased at convenience stores. PayPal had captured the electronic minority, but that minority represented the company’s entire economic foundation, providing the volume necessary to keep the data centers running and the fraud algorithms learning.
This dependency created a strategic captivity that threatened the company’s survival even as it fueled its growth. PayPal needed eBay’s transaction volume to achieve the scale necessary to cover its fixed infrastructure costs. Yet remaining tethered to the marketplace risked permanent classification as merely a feature of eBay rather than a standalone payments network. If eBay executives decided to cut off API access, mandate the use of their in-house alternative Billpoint, or simply alter the terms of engagement to extract more value, PayPal would lose two-thirds of its volume instantly. The fixed costs would remain, lease agreements on server facilities, salaries for fraud analysts, regulatory compliance overhead, but the revenue to cover them would vanish, collapsing the thin-margin model into insolvency.
The gamble rested on a fragile assumption: that users who adopted PayPal to settle auction transactions would organically transition to general peer-to-peer money transfer. The company bet that the network density achieved within eBay’s walls would create sufficient switching costs that users would maintain their accounts, and their balances, for transactions outside the marketplace. But this assumed that eBay would tolerate the leakage of economic activity away from their platform, and that Billpoint would fail to achieve product parity. Both assumptions were precarious. eBay viewed electronic payments not as an independent industry but as a critical feature of their marketplace infrastructure, a necessary lubricant for commerce that they should control. They acquired PayPal in July 2002 for $1.5 billion, a mere five months after the IPO, precisely because they recognized that controlling the payment flow was essential to controlling the market. The acquisition solved PayPal’s immediate liquidity concerns but intensified the captivity problem. Now the platform owner and the payment provider were the same entity, and the tension between platform integration and network independence became an internal civil war fought through budget allocations and product roadmaps.
The Fraud Engine and the Velocity Constraint
Most analyses of PayPal’s defensibility focus on the network effects, the brand recognition, or the first-mover advantage in online payments. These matter, but they obscure the true structural barrier to entry and the central tension of the business model: the company had to grow fast enough to cover fixed costs, but not so fast that fraud consumed the thin margins. This scale
velocity versus fraud cost containment tension defined the operational reality of the business. Every new user brought potential volume but also potential liability. Fraud in electronic payments does not scale linearly; it compounds. A stolen credit card tested successfully against a merchant account opens the door to systematic extraction. At the thin margins PayPal operated on, a take rate of roughly two to three percent on most transactions, a fraud loss rate exceeding one percent threatened to convert the entire enterprise into a mechanism for subsidizing criminal activity. The mathematics were unforgiving. If PayPal processed ten billion dollars in annual volume and fraud consumed even two percent of that flow, the two hundred million dollars in losses would swamp the gross margin, leaving nothing to cover the fixed infrastructure costs that ran into the hundreds of millions annually.
The company responded by building a proprietary fraud detection infrastructure that became its most underappreciated structural advantage. This was not merely a compliance function or a risk management checkbox; it was the core technical moat that permitted the business model to exist at all. PayPal’s engineering teams developed machine learning systems capable of analyzing thousands of data points in milliseconds, device fingerprints, behavioral biometrics, transaction velocity patterns, network graph analysis of sender-receiver relationships, to identify and block fraudulent transactions before they cleared. The sophistication of these systems created a high-fixed-cost barrier that new entrants could not cross without bearing equivalent fraud losses during their learning curve. A competitor attempting to replicate PayPal’s user experience would face a brutal education: fraudsters probe new payment platforms systematically, exploiting the lack of historical data to train detection algorithms. The losses during this ramp-up period are catastrophic, explaining why, as one observer noted, many competing services from that era “have bodies buried all over” from uncontrolled fraud.
This technical infrastructure carried secondary economic consequences beyond loss prevention. The fraud detection capabilities developed at PayPal directly enabled the creation of Palantir, the data analytics company founded by Peter Thiel and former PayPal engineers who had built the initial risk management systems. The intellectual capital embedded in PayPal’s fraud algorithms represented a spin-off value that exceeded the immediate utility of loss prevention. More importantly, the fraud engine allowed PayPal to maintain its regulatory classification as a money transfer service rather than a bank. By demonstrating sophisticated internal controls over stored value and transaction monitoring, the company could argue to regulators that it posed no systemic risk requiring banking licensure. This classification arbitrage was not a legal technicality; it was fundamental to the margin structure. Banking regulation would have imposed reserve requirements, capital adequacy ratios, and compliance costs that the thin-take-rate, high-volume model could not support. Reclassification would have broken the economics, transforming a viable network utility into an unprofitable regulated entity.
The Regulatory Tightrope and the Margin Architecture
PayPal’s existence depended on maintaining a specific legal fiction: that it was not a bank, not a money transmitter in the traditional regulatory sense, but something novel that required lighter touch supervision. This was not accidental oversight but active architectural design. The company stored value in user accounts, balances that could be sent, received, and held indefinitely, functionally behaving like demand deposits. Yet it operated without Federal Reserve oversight, without FDIC insurance requirements, without the capital reserves that banking statutes mandate. This regulatory arbitrage created the margin structure that made the business viable.
The tension between classification arbitrage and compliance manifested in constant legal and technical maneuvering. Every product decision risked triggering reclassification. Offering interest on stored balances would suggest banking activity. Extending credit directly rather than through partner banks would engage lending regulations. Holding user funds in pooled accounts rather than individual fiduciary accounts risked different regulatory treatment across state and federal jurisdictions. PayPal’s legal team operated as a shadow product organization, vetoing features that might cross the invisible line between money transfer service and depository institution.
This fragility remains unresolved. The company’s regulatory status exists at the pleasure of interpretation. Any shift in banking statutes, any reinterpretation of stored value regulations following a major fraud incident or systemic shock, could impose compliance costs that the model cannot absorb. The bear case for PayPal has always rested here: not on competitive threats from other payment apps or technological obsolescence, but on the possibility that regulators eventually recognize the company as a shadow banking system and apply the full weight of financial institution oversight. The thin margins that make the business attractive to users, low transaction fees subsidized by float income and scale efficiency, cannot survive the cost structure of regulated banking. This is the fragile long-term assumption embedded in the equity: that the legal classification will persist indefinitely despite the company’s growth to a scale that once defined systemically important financial institutions.
The Exodus of the Mafia and the Depreciation of Human Capital
When eBay completed the acquisition in July 2002, the transaction solved PayPal’s immediate liquidity crisis but introduced a cultural conflict that functioned as a direct economic variable. eBay operated as a corporate entity with defined hierarchies, quarterly planning cycles, and integration playbooks developed through the assimilation of prior acquisitions. PayPal operated as a startup that had survived near-death experiences through rapid iteration, product improvisation, and the obsessive focus on viral growth metrics that characterized the PayPal Mafia: he cohort of founders and early engineers who would later seed Tesla, LinkedIn, Palantir, YouTube, and Yelp.
The cultural autonomy versus corporate integration tension manifested in budget allocations and reporting structures, but its economic impact appeared in the departure of talent. Within four years of the acquisition, over fifty percent of PayPal employees had left the company, including all founders from both Confinity and X.com. This was not ordinary post-acquisition turnover; it represented the elimination of the iterative product capability that had created the original product-market fit. The founders understood the recursive network density model intuitively because they had built it through desperate experimentation. Their departure transferred that institutional knowledge to external ventures while leaving eBay with a maintenance operation rather than an innovation engine.
Peter Thiel, Max Levchin, Luke Nosek, David Sacks, Elon Musk, hese names represented more than leadership; they embodied the specific technical and strategic competence required to balance scale velocity against fraud containment. Their exit was not a soft factor of “culture fit” but a hard economic loss equivalent to writing off a major fixed asset. When Thiel departed, he took with him the conceptual framework for fraud detection that would become Palantir. When Sacks left, he took the product intuition for viral mechanics. The remaining organization could process transactions and maintain infrastructure, but the capacity for the kind of zero-to-one innovation that had built the recursive network in the first place evaporated.
eBay faced a choice that every platform owner acquiring a network business must confront: impose corporate governance and risk losing the human capital that created the value, or grant operational autonomy and accept that the acquired entity may optimize for its own network growth rather than the platform’s interests. eBay chose integration, and the economic consequence was a decade of underinvestment in PayPal’s standalone capabilities. The payment system remained tethered to the auction marketplace, starved of the resources necessary to build the general peer-to-peer infrastructure that would eventually justify the 2015 spinoff valuation. The talent exodus froze product development in amber; PayPal survived the eBay years not through innovation but through the compounding network effects already achieved, riding the density accumulated before the acquisition while slowly bleeding the technical capacity to defend it.
Viral Mechanics and the Illusion of Zero Cost
The density threshold required user acquisition at speeds impossible through traditional marketing. PayPal could not afford the customer acquisition costs that banks or credit card companies deployed; the lifetime value of a user, given the thin take rates, did not support expensive advertising or sales forces. The solution lay in viral distribution mechanics that lowered the marginal cost of new nodes to near zero while accelerating the recursive network effects.
The company pioneered the “give five dollars, get five dollars” referral model, a mechanism that converted existing users into a distributed sales force. Unlike traditional affiliate programs that paid for impressions or clicks, PayPal only paid for verified account creation and initial transaction completion. This aligned incentives perfectly: users promoted the service because they received immediate cash, and PayPal only incurred cost when the new user became economically active. The model created exponential growth curves that looked like zero-cost acquisition but actually represented a precisely calibrated cost of capital. PayPal was essentially borrowing growth from its users, paying interest in the form of referral bounties, and banking on the network density to generate sufficient transaction volume to repay the loan with operational cash flow.
Simultaneously, the company deployed HTML embed buttons that allowed any merchant or individual to add payment functionality to websites with a single line of code. This embeddability transformed every blog, every personal homepage, every small business website into a distribution node for the network. The technical simplicity masked strategic sophistication: by making integration trivial, PayPal ensured that merchants brought their own customers into the network, who then became potential senders in their own right, who then became potential receivers. The viral loop closed recursively.
However, these mechanics created their own fragilities. The referral model assumed that the lifetime value of acquired users would exceed the upfront bounty. If users signed up, collected the five dollars, and transacted only once, or if they gamed the system through multiple account creation, the unit economics collapsed. The embeddability assumed that merchants would police their own fraud exposure; when they did not, PayPal absorbed the losses. The viral growth that solved the density threshold problem introduced quality control problems that threatened the fraud containment systems. The company was simultaneously trying to grow as fast as possible while maintaining the fraud rate below one percent, a balancing act that required the machine learning infrastructure to improve faster than the fraudsters could adapt.
The Billpoint Shadow and the Platform Captivity Risk
The specific mechanics of PayPal’s dependency on eBay become clearest in the examination of Billpoint, eBay’s in-house payment solution launched in direct competition with PayPal. At the time of acquisition, eBay viewed PayPal not as a strategic investment in the future of money but as a feature acquisition: necessary patch for their marketplace’s friction points. They had already invested in Billpoint, a joint venture with Wells Fargo designed to provide electronic payment services exclusive to eBay users. The platform captivity risk was not theoretical; it was embodied in a competing product that eBay controlled and preferred.
Billpoint offered integration advantages that PayPal could not match. It appeared as the default payment option in eBay listings. It processed transactions without requiring buyers to leave the eBay ecosystem. It settled funds directly into eBay’s managed accounts. From eBay’s perspective, Billpoint represented the ideal: complete control over the payment flow, full visibility into transaction data, no revenue leakage to an external provider, and no risk that the payment provider would facilitate off-platform commerce. If Billpoint had achieved product parity with PayPal, if it had matched the fraud detection capabilities, the ease of account setup, the viral distribution, the platform owner would have had no incentive to maintain PayPal’s privileged position.
PayPal survived this existential threat because Billpoint failed to clear the density threshold. Lacking the recursive network effects, unable to leverage eBay users for off-platform growth, unable to convert senders into receivers outside the auction context, Billpoint remained a single-use-case feature rather than a growing network. Its fraud detection lagged because it lacked the transaction volume necessary to train effective algorithms. Its user experience suffered because eBay treated it as a cost center rather than a growth engine. The sixty percent of eBay transactions still settling by check in 2002 represented Billpoint’s failure to capture the electronic market, not PayPal’s inevitability.
Yet the shadow of Billpoint shaped PayPal’s strategic behavior during the eBay years. The threat of being replaced, of having API access restricted, of seeing the “Pay with PayPal” button demoted in search results or removed from preferred placement, forced PayPal to accept integration terms that starved standalone development. The company could not invest aggressively in mobile infrastructure, in peer-to-peer social payments, in merchant services outside the eBay ecosystem, because eBay’s corporate parent viewed such investments as competing with the core marketplace. PayPal remained profitable during these years, eBay’s transaction volume provided the density necessary to cover fixed costs, but it became a captive utility, a regulated monopoly within the auction platform rather than an independent network.
This captivity created a specific type of strategic myopia. PayPal’s management focused on optimizing the eBay experience, reducing checkout friction, lowering fees for high-volume sellers, integrating with eBay’s dispute resolution systems, while neglecting the general payments infrastructure that would eventually justify a standalone valuation. The assumption that users would organically transition from auction payments to general money transfer remained untested because eBay prevented PayPal from aggressively pursuing off-platform use cases. The network density achieved on eBay created switching costs, yes, but it also created switching inertia. Users maintained PayPal accounts because they needed them for eBay; they did not necessarily develop habits of using PayPal for rent payments, dinner splits, or freelance services. The recursive network remained theoretically extensible but practically constrained.
The Spinoff and the Valuation Inversion
When eBay spun off PayPal in July 2015, the market capitalization dynamics revealed the economic consequences of the cultural and strategic tensions that had played out over thirteen years. At separation, PayPal traded at an initial value approximately 1.5 times eBay’s market capitalization: oughly $46 billion versus $30 billion for the marketplace that had once acquired it for $1.5 billion. The payment network had exceeded the value of the platform that had provided its density foundation.
This inversion validated the standalone network thesis but also exposed the opportunity cost of the captive years. During the decade of eBay ownership, PayPal had grown primarily through the natural expansion of e-commerce rather than through the aggressive pursuit of network density in general payments. The spinoff valuation suggested that had PayPal maintained its cultural autonomy, had it invested in the fraud detection and viral mechanics that required freedom from eBay’s integration playbook, the network might have achieved greater density earlier. Instead, the company spent thirteen years optimizing for a single use case while competitors, Stripe in merchant services, Venmo (later acquired by PayPal itself) in peer-to-peer transfers, Square in mobile point-of-sale, built the standalone infrastructure that PayPal might have developed.
The spinoff also clarified the regulatory and fraud containment architecture as the true sources of value. Investors valued PayPal not merely as a transaction processor but as a trusted escrow service with proprietary risk management capabilities. The fraud detection moat, built during the desperate early years and maintained through the eBay period, represented the durable asset that new entrants could not replicate. The recursive network effects provided the volume, but the fraud infrastructure provided the viability. Without the ability to maintain loss rates below take rates at scale, the network density would have been a liability rather than an asset.
The Recursive Trap
The bear case for PayPal rests on the fragility of the assumption that network density creates permanent switching costs. The model assumes that users who adopt PayPal for one use case, originally eBay auctions, now perhaps online merchant checkout, will organically expand their usage to other transaction types, increasing the platform’s utility and justifying the infrastructure costs. But this assumption conflates account existence with active network participation.
Users maintain PayPal accounts in dormancy. The density exists on paper, hundreds of millions of accounts, but the recursive utility depends on active transaction graphs. If users maintain PayPal primarily as a legacy payment method for specific merchants while conducting daily financial interactions through banking apps, Apple Pay, or other closed-loop systems, the network becomes brittle. The density that provided protection against Billpoint in 2002 does not necessarily protect against embedded finance in 2024, where every app contains its own payment functionality and the general-purpose wallet becomes redundant.
Furthermore, the fraud detection advantage, while durable, faces erosion through the democratization of machine learning tools. The high fixed costs that once protected PayPal, millions in infrastructure investment to train fraud models, have fallen as cloud computing and open-source algorithms lower the barrier to entry. A new entrant today faces less brutal fraud losses during ramp-up because they can rent sophisticated detection capabilities from third-party providers rather than building proprietary systems. The moat narrows.
The structural constraint remains unchanged: the business requires massive volume to cover fixed costs, but the viral mechanics that generated that volume depended on a specific internet era: n era of open platforms, email-based viral loops, and the absence of embedded payment options in social media and messaging apps. PayPal’s recursive network was built for a web of separate sites; it is not obviously optimized for a web of integrated applications. The density threshold that once represented an insurmountable advantage may become a legacy cost structure in a world where payments fragment into platform-specific closed loops.
The company survives today on the compounding momentum of decisions made twenty years ago: he referral bonuses, the embeddable buttons, the fraud algorithms developed by founders who departed within four years of the acquisition. Whether that momentum constitutes durable advantage or structural inertia remains the open question. The recursive network model works until it doesn’t, until the density becomes irrelevant to how money actually moves. The thin margins that required the density in the first place leave little room for error if the density begins to dissipate.
The dormancy problem manifests not merely as inactive accounts but as a specific liability on the balance sheet. Every dormant account represents stored value, unds sitting in PayPal’s float that generate investment income, yes, but also regulatory exposure, customer service infrastructure, and compliance monitoring costs that do not scale down to zero simply because the user no longer transacts. The recursive network model assumes that density, once achieved, maintains itself through utility. In practice, maintaining the infrastructure for three hundred million accounts costs substantially the same whether they transact daily or annually. When the active transaction graph contracts to a subset of merchant categories, online gaming, cross-border remittances, specific e-commerce verticals where PayPal retained preference, the fixed costs remain distributed across a shrinking revenue base. The density threshold becomes a fixed cost trap.
The viral mechanics that originally solved the density problem belonged to a specific architectural era of the internet: the web of discrete sites, hyperlinks, and email-based identity verification. PayPal’s growth relied on the embeddable button placed on independent merchant sites, driving users back to PayPal’s centralized interface to authenticate and complete transactions. This created a hub-and-spoke model where PayPal remained the necessary intermediary. The contemporary mobile environment operates on inverse principles. Super-apps and embedded finance integrate payment functionality directly into the merchant or social experience, eliminating the redirect to a centralized wallet. When a user splits a dinner bill within a messaging thread or checks out within an Instagram storefront without leaving the app, the recursive network effects accrue to the platform operator, not to the payment utility. PayPal’s density was built for a topology of separation; modern commerce rewards topologies of integration.
The human capital depreciation that occurred during the eBay years compounded this architectural obsolescence. The fraud detection systems that provided the essential moat were not merely technical artifacts but living institutional capabilities requiring continuous iteration. The machine learning models built by Thiel, Levchin, and the early engineering cohort were trained on specific patterns of early internet commerce, auction fraud, digital goods non-delivery, account takeovers via password reuse. As commerce migrated to mobile and new fraud vectors emerged, synthetic identity theft, device spoofing at scale, social engineering through platform messaging: he absence of the original architects meant slower adaptation. The teams that replaced them were competent administrators of existing infrastructure but lacked the specific combative ingenuity required to rebuild detection systems for fundamentally different threat environments. A fraud moat, unlike a patent, requires constant excavation. When the diggers depart, the moat silts up.
Consider the specific case of account takeover defenses in the transition from web to mobile. The early PayPal systems relied heavily on email verification loops and cookie-based device fingerprinting, ffective when users accessed the service from desktop browsers with persistent IP addresses. Mobile devices introduced fragmented session management, shared devices among family members, and IP addresses that shifted with network connectivity. Adapting the fraud models to distinguish legitimate mobile usage from coordinated attacks required the kind of zero-to-one technical innovation that the post-acquisition organization struggled to execute. The result was conservative risk thresholds that generated false positives, freezing legitimate accounts and degrading user experience, or permissive thresholds that admitted fraud losses precisely when competitors with native mobile infrastructure were optimizing their own detection. The fixed cost advantage of the fraud infrastructure became a fixed cost burden when the infrastructure required fundamental rebuilding rather than incremental maintenance.
The regulatory arbitrage that preserved the margin structure faces similar temporal erosion. The classification as a money transfer service rather than a bank depended on the limited functionality of early PayPal: ssentially a digital escrow for specific transactions. As the company added credit products, savings features, cryptocurrency trading, and merchant lending, it approached the functional equivalence of a banking institution without the corresponding regulatory capital. Each product expansion increased the probability of reclassification, yet the thin margins required product expansion to generate revenue diversification. The company navigates a narrowing corridor: add features to monetize the dormant base and risk banking regulation; remain a simple wallet and watch the active transaction graph migrate to full-service competitors.
The specific mechanics of the 2015 spinoff illustrate this tension. eBay retained a five-year operating agreement requiring PayPal to process a significant percentage of eBay’s payments, providing transitional revenue stability but preventing immediate strategic pivoting. During those five years, mobile payment adoption accelerated, embedded finance matured, and the window for PayPal to establish standalone density outside the auction context narrowed. The spinoff solved the cultural captivity problem too late to address the architectural transition. PayPal received its independence at precisely the moment when the independent wallet model was being subsumed by integrated financial operating systems.
The recursive network’s final fragility lies in the asymmetry of defection. In a dense two-sided network, merchants and consumers theoretically create mutual stickiness. In practice, merchant defection cascades faster than consumer attachment. When a major retailer removes PayPal as a checkout option, citing lower fees from direct bank integration or superior conversion from Apple Pay, the consumer does not abandon the retailer; they simply update their payment method. The network density that appeared to create switching costs actually concentrated risk in the merchant relationship. Lose the merchant, and the consumer follows, not the reverse. The density becomes a liability when the maintenance costs of the infrastructure exceed the marginal retention value of specific merchant relationships.
This asymmetry exposes the fundamental economic reality obscured by the network effects narrative. PayPal’s value creation depended on becoming the default: he wallet that required no thought. But defaults are context-dependent, and the contexts multiplied. The recursive density that justified the high fixed costs required ubiquity; ubiquity required accommodating every use case; accommodating every use case required regulatory and technical complexity that eroded the thin margins. The business that began as a viral loop for auction payments ends as a legacy infrastructure provider maintaining dormant accounts for a shrinking subset of transaction types, still paying for the data centers that made sense at scale but became millstones when the scale fragmented.