Risk-free stack

For seventy years, the dollar and U.S. Treasury bonds have been a single transaction. Owning one was equivalent to owning the other. A central bank seeking to safeguard U.S. security bought U.S. Treasury bonds. By buying the bonds, it held the U.S. currency. Thus, the privileges of being the world's currency and the world's safest asset were merged into a single financial instrument with a single yield. That weld is about to break. Not in crisis headlines, not in defaults, not in dramatic moments of one asset replacing another. It's erupting in the only place where such an event would foreshadow it: the price the world pays for security far exceeding its returns. The premium for holding the dollar remains strong. The premium for holding long-term U.S. Treasury bonds has narrowed significantly, and the premium for long-term bonds has even turned negative. This is not a prediction, but the conclusion of a paper published in December 2025 by Weinhardt, Ritter Girati, and Jesse Schreger. The paper was published as Federal Reserve International Finance Discussion Paper 1427 and National Bureau of Economic Research Working Paper 35000. They distinguished monetary convenience from bond convenience by deviating from hedge parity and documented the decoupling between the two. The convenience of the dollar remains strong. The convenience of U.S. Treasury bonds has declined significantly, even turning negative, especially for medium- and long-term bonds. The world is still willing to transact in dollars, but no longer willing to store decades of value in the duration of U.S. Treasury bonds under the old conditions. This means a lot, and the framework that maps it is not the framework carried by most allocators. Risk-free assets are never singular. They are originally a combination of services, and there just happens to be a financial instrument that can provide all those services simultaneously. Today, this suite of services is fragmented into multiple layers, each provided by different financial instruments, each repricing at its own pace, and each bearing its own custody, jurisdiction, settlement, and political risks. The question that once had only one answer—what is a risk-free asset—now has several, and these answers no longer agree. First, a boundary must be drawn, because the entire argument depends on this boundary. Geopolitical risk has already been priced in, which is nothing new. The International Monetary Fund priced it in its April 2025 Global Financial Stability Report, using news-based geopolitical risk indicators, sanctions variables, sovereign credit spreads, and asset yields to show that major shocks, especially military conflicts, bring a sustained and measurable premium, and calculated the so-called geopolitical risk beta coefficient. The European Central Bank and the European Committee on Systemic Risk have also established a fragmented monitoring framework. Academic literature has long considered sensitivity to geopolitical risks as a pricing factor. These are not mine. Anyone claiming to have found that political factors affect prices is either uninformed or dumping their goods. The contribution here is not in proposing new factors, but in decomposition. Observing the disintegration of the overall structure, pointing out the separated levels, and raising a question that existing research has not yet addressed: Is the next effective measure of security structural or reactive? Integration, not revelation. Where the framing argument holds true, it will be stated truthfully; where the framing argument becomes a research project, it will also be explained unambiguously. For readers who are simultaneously opening an IMF report in another window, this sentence lays the foundation for the rest of the article. So, here it is. And how it was damaged. The meaning of security in the past. For much of the postwar period, the U.S. Treasury's instrument of tradable debt served five functions for the global system so much that almost no one thought to separate them. It was reserves, assets held by central banks to anchor national currencies and store national wealth in the world's unit of account. It was collateral, a common pledge in the repurchase market and eligible Tier 1 liquid assets under banking liquidity rules. It was cash, Treasury bonds, where corporations, funds, money market instruments, and dollar institutions stored and operated liquidity. It was once a store of value, long-term bonds purchased by pension funds, insurance companies, and sovereign wealth funds to ensure decades of security. Below these four lies settlement, the transfer of asset ownership through a clearing system, which is widely believed to remain open indefinitely. The brilliance of this arrangement—what a French finance minister once called America's "excessive privilege"—lies in the fact that these five bonds are bundled together at a uniform price. You don't need to choose. By purchasing U.S. Treasury bonds, you gain reserve status, collateral utility, cash equivalents, duration, and final settlement rights from the same issuer all at once, with a uniform yield. The convenience yield measures the value of this arrangement. For decades, its value has been enormous. Today, this bundled sale is crumbling. Services are separating. Different financial instruments are playing different roles. Tools that once provided all five services are showing remarkable strength in some areas and significant weakness in others, a typical characteristic of bundled products being gradually broken down and repriced. Layer by layer, they are being dismantled. The long end has lost its immunity. Starting with where the pressure is undeniable and the cause irrefutable, is the storage of long-term value. The reason is fiscal arithmetic, and not a subtle one. The Congressional Budget Office, in its budget and economic projections covering 2026 through 2036, projects that the federal deficit will reach $1.9 trillion this fiscal year, equivalent to 5.8% of output, and will expand to $3.1 trillion, or 6.7%, by 2036. Publicly held debt will climb from 101% of GDP to 120%, surpassing the post-World War II record of 106%. The deficit is also changing its composition in the worst possible direction. Net interest (i.e., the cost of debt) as a percentage of output will rise from 3.3% in 2026 to 4.6% in 2036, accounting for nearly one-fifth of total federal spending. The office notes that net interest as a percentage of GDP will exceed 3.2% annually during the forecast period, the highest level since at least 1940. This baseline itself is now subject to legal uncertainty due to the Supreme Court's February 2026 ruling that the government's emergency tariffs were invalid, leading to controversy over a large portion of the expected tariff revenue. The fiscal problems do not depend on the specific tariff adjustments; debt service mechanisms are already in place. This is the engine. A government burdened with a structural deficit, while interest payments continue to accumulate, forcing it to continuously issue debt to a world with limited demand for this single instrument. Rating agencies have documented this. Moody's downgraded the US credit rating from Aaa to Aa1 in May 2025, causing all three major rating agencies to remove the US from their top-rated list, citing rising debt, persistent deficits, and increasing interest costs. S&P Global Ratings adjusted its rating in 2011, Fitch in 2023, and the fourth-largest rating agency, Scope, in October 2025. Du, Kiraty, and Schreiger are responsible for translating arithmetic into prices. The decline in the accessibility of US Treasury bonds stems from supply issues. The relative abundance of US Treasury bonds compared to other developed sovereign bonds is eroding their premium. The convenience of the US dollar remains strong, but the convenience of bonds has diminished, even turning negative in medium- and long-term bonds. The situation in Europe is quite the opposite. The European Central Bank, in its June 2026 assessment report, noted that the convenience yield on German government bonds is rising due to strong global demand for high-rated safe-haven Eurozone assets. This distinction is significant. The world has not abandoned the dollar, but rather is reassessing the duration risk of sovereign nations that issue dollars. This is a crack that most portfolios cannot foresee. A dollar shortage may coexist with inadequate duration digestion in government bonds. Therefore, the first layer of structure begins to separate. Long-term assets are gradually becoming ordinary risk assets, their pricing no longer determined by blind risk aversion, but by supply, term premium, and fiscal creditworthiness. This is not a default, but a repricing. What was considered safe for decades is no longer seen as safe after a long period of development and maturation. The front end has secured exclusive offers. Now, the simple story that the world is losing confidence in US debt is beginning to crumble. While the long end is weakening, the short end is strengthening. The source is surprisingly in the most unlikely corner of modern finance: dollar-backed stablecoins. Privately issued tokens promising redemption at face value have become major buyers of Treasury bonds. In July 2025, Congress passed legislation enshrining this demand in law. The GENIUS Act took effect on July 18, 2025. The bill restricts bond issuance to authorized issuers, requires a minimum 1:1 ratio of cash and certain liquid asset reserves, including short-term government bonds and qualified repurchase agreements, severely limits the reuse of these reserves, and mandates monthly public disclosure of reserve composition. The result is not that cryptocurrencies are escaping national regulation, but rather a regulated front-end distribution channel for government bonds. Every dollar invested in compliant payment stablecoins is pushed into a narrow collateral market. Front-end traders can find pre-set buyers, while back-end traders cannot. This figure is real and must be stated truthfully, not exaggerated, because this is precisely the figure most easily exaggerated and then overturned. In their working paper on stablecoins and safe-haven asset prices (revised February 2026), the Bank for International Settlements (BIS) used daily data from 2021 to 2025 to find that an inflow of approximately $3.5 billion (equivalent to two standard deviations) would reduce the yield on three-month Treasury bonds by 2.5 to 3.5 basis points within ten days. This impact depends on market conditions: when Treasury supply is ample, the effect is not statistically significant; while when Treasury supply is scarce, the effect rises to 5 to 8 basis points. This impact is primarily concentrated on short-term Treasury bonds, with limited or almost no spillover effects on long-term Treasury bonds. Tether, the largest issuer, contributes the most, followed by Circle. According to the issuers' own reserve reports, by the end of 2025, major dollar-backed tokens will hold over $270 billion, of which approximately $153 billion will be in Treasury bonds, and approximately 33 billion Treasury bonds have been purchased in the past year. A weak force on the periphery, not a torrent. But the reality is quite the opposite of the long-term trend. The cash layer of the dollar system is deepening due to private digital demand, while the duration layer is shrinking due to fiscal supply. Treasury bonds and general bonds, once considered different maturities of the same asset, are now fragmented by various forces. This is being broken down within the Treasury yield curve. There is a deeper irony here that underpins the following argument. Stablecoins strengthen the dollar's position, but they do not free holders from the dollar system's constraints. It is this legislation mandating stablecoin reserves that places issuers under a regulatory framework and requires payment-based stablecoin systems to comply with legal instructions, including instructions to seize, freeze, destroy, or prevent the transfer of specific tokens. The digital dollar is not anti-sovereign currency. It is a way for sovereign currencies to be programmably extended through private channels. This makes it a powerful tool for the dollar's power, but also a weak one for escaping it. Central banks worried about their dollar reserves being frozen cannot eliminate this concern even when holding dollar claims, which are more easily frozen. This is a clue to the next layer. Reserve insurance is transforming, not collapsing. The reserve tier, the assets held by official institutions to anchor the currency and store national wealth, is an area where security politics are becoming increasingly clear. In early 2022, the G7 froze approximately $300 billion in reserves held by the Russian central bank. This marks the first time since the modern era of reserves that officials have witnessed a major nation's core security asset become unusable, not due to default, but because custodians and clearing systems, under political directives, have refused access to the funds. The asset still exists. Its owner simply cannot touch it. Under adversarial conditions, the distinction between credit and availability has shifted from a footnote to a core issue. Let's begin by refuting exaggerated arguments, as these are compelling. The dollar's dominance has not collapsed. The dollar remains the largest reserve currency, by a wide margin, accounting for nearly 57% of allocated reserves according to the latest official data. The institutions that compiled these data believe that recent exchange rate fluctuations are primarily due to exchange rate valuation rather than deliberate selling. The dollar still dominates the vast majority of currency transactions. A Federal Reserve study found that roughly three-quarters of official safe-haven asset holders are governments allied with the U.S. military, leaving them with little reason and little room for maneuver. Both conditions hold true simultaneously. The dollar can maintain its dominance in financing and trading, while the official sector can diversify in terms of insurance. Different levels require different tools. Observe the movements in the insurance layer. The World Gold Council reported that central banks purchased a total of 863 tons of gold in 2025, lower than the more than 1,000 tons per year from 2022 to 2024, but far higher than the 473 tons per year in the decade before the freeze. In 2026, gold purchases accelerated again, with 244 tons purchased in the first quarter, a 17% increase quarter-on-quarter and a 3% increase year-on-year, with Poland and Uzbekistan being the main buyers. The association's 2025 survey of central banks is its largest to date, covering 73 central banks. The survey found that 95% of the surveyed central banks expect global official gold reserves to increase in the coming year, and a record 43% expect their own reserves to increase as well. Discipline is more important than drama. The European Central Bank reported that by the end of 2025, gold will account for 27% of global official reserves by market capitalization, surpassing the euro (15%) and US Treasury bonds (22%). This figure is striking, but price alone is misleading, as it primarily reflects price movements. If recalculated based on the gold price at the end of 2023, excluding upward price movements, US Treasury bonds would far outweigh the gold. Gold is being steadily accumulated for obvious reasons. Gold has not replaced US Treasury bonds as the core liquid asset of the official sector; to believe that gold has replaced US Treasury bonds is a misinterpretation of price charts as a shift in asset structure. Gold also has its limits, which are rarely encountered by its enthusiasts. Gold is more effective than currency exchange in dealing with currency freezes. Russia is a case in point. For years, Russia has been reducing its dollar holdings and stockpiling gold. When a currency freeze occurs, this gold, though stored in Russian vaults and difficult to freeze, is almost unusable for purchasing the currencies of sanctioned countries that Russia needs, except in cases of barter with them. Available controls are not a single property. An asset may be both difficult to seize and difficult to dispose of, or it may be easy to dispose of but easy to seize. The challenge for reserve managers is how to hold both types of assets simultaneously. No single instrument can satisfy both needs at the same time. Deconstruction at the national level is necessary. Behind the news about reserve quotas, a quiet shift is occurring in the group of U.S. Treasury buyers. Data from the U.S. Treasury Department's International Capital Markets shows that net foreign capital inflows in March 2026 totaled $150.7 billion, but the key lies in the composition of these funds. Foreign private investors bought $162.1 billion, while official institutions were net sellers of $11.4 billion. Regarding long-term securities, foreign private investors bought $111.4 billion, while foreign official institutions sold $14.9 billion. The Treasury Department also cautions that custody data cannot fully and accurately reflect ultimate ownership. The Federal Reserve's custody accounts reflect the same situation. As of the end of May 2026, the Federal Reserve held approximately $2.69 trillion in Treasury securities for foreign official institutions and international institutions, a decrease of approximately $225 billion from the previous year, while its total custody holdings were close to $2.97 trillion, a decrease of nearly $290 billion itself. The character of marginal buyers is changing. From price-insensitive officials managing fixed exchange rates to price-sensitive private investors, money market instruments, stablecoin issuers, hedge funds, and bank balance sheets. But this doesn't mean no one is buying American products. It means the motivations of buyers have changed. And motivations determine people's behavior under pressure. Even so, there's a concern about over-interpreting, and honesty requires us to do so. A 2025 study by the Federal Reserve found that capital data significantly underestimates foreign assets flowing through offshore financial centers, underestimating by approximately $1.4 trillion in the Cayman Islands alone. This means that a large portion of documented private demand may be leveraged or intermediary flows rather than the slow-moving funds of true reserve managers. But this doesn't diminish the shift in asset structure. It will only make, not weaken, the front end's dependence on this demand more vulnerable. Channels are valuable assets. The bottom layer, and the least noticed layer, of the old package is: settlements. It assumes not only that the asset can be profitable, but also that the channels through which the asset operates remain unobstructed. Freezing foreign exchange reserves shattered this assumption, and the effects are now global. ECB officials are increasingly inclined to view payments as sovereign rather than infrastructure. In April 2026, an ECB Governing Council member pointed out that Europe's dependence on payment infrastructure outside the continent is a strategic vulnerability. He noted that a large portion of bank card transactions in the Eurozone rely on non-European payment systems and viewed the interconnection of the digital euro and instant payment systems as a defense against external influence and decoupling. The same logic is reflected in the IMF's January 2026 report on cyber risks, which views the concentration of power in a few cloud service providers as a systemic problem. The validity of payment requests depends on the infrastructure beneath them, which itself has owners, jurisdictions, and bottlenecks. These are all influenced by political factors. The same competition is playing out on a larger scale in the computing field. The most important strategic asset of the next decade is not just financial instruments, but computing power—including chips, data centers, energy, water resources, grid interconnection, software stacks, and licenses for training and running artificial intelligence. Today, every major country views it as critical infrastructure, with its control objective being security, not commerce. S&P Global defined computing sovereignty in May 2026 as a structural risk encompassing hardware, software, jurisdiction, and operational control, noting that suppliers or governments can suspend or revoke licenses for advanced chips. This is not just theoretical. The U.S. authorized the export of up to the equivalent of 35,000 Nvidia advanced chips to state-owned entities in Saudi Arabia and the United Arab Emirates by the end of 2025, but attached stringent security and reporting conditions. Chips come with security boundaries from the moment they leave the factory: licenses, reporting, end-use restrictions, and the continuing discretion of the exporting sovereign state. Computing is the purest example of asset availability control, which depends on the continued permission of the state, suppliers, energy systems, and regulators—the same attributes found in frozen accounts by reserve managers, now etched onto silicon wafers. Computing is being financialized. In March 2026, AI cloud service provider CoreWeave secured an $8.5 billion deferred drawdown term loan to expand its platform. The initial borrowing was approximately $7.5 billion, with subsequent borrowing increasing as its data center assets become operational. The loan, maturing in March 2032, was designed and underwritten by Morgan Stanley and Mitsubishi UFJ Financial Group (MUFG), with Goldman Sachs and JPMorgan Chase acting as lead arrangers, and Blackstone Credit & Insurance providing guarantees. It received a Moody's A3 investment-grade rating and is the first investment-grade financing project secured by high-performance computing infrastructure and its associated customer contracts. Capital is shifting towards the physical level of digital sovereignty. However, this is not without risk. It is not a new treasury. Credit ratings depend on customer contracts, utilization rates, electricity, depreciation, export licenses, and counterparty quality. The strength of core customers can mask the vulnerability of peripheral customers. Investment-grade ratings borrow the strength of a few dominant platform customers and lend it to hardware that is depreciating every quarter. The counter-argument presented here is, in my opinion, the most effective rebuttal to this lazy approach to argumentation. Nations are not passively relinquishing infrastructure and computing power to corporations; rather, they are attempting to regain control of these resources. The Brookings Institution pointed out in February 2026 that almost no country can structurally achieve complete AI sovereignty, and a more realistic model is controlled interdependence: governments selectively develop domestic capabilities while relying on supply chains they cannot replicate. Canada, the EU, and Gulf states have all launched autonomous computing strategies. This story is not directed at any particular institution. It is a struggle. Nations are using export controls, industrial policies, licensing regimes, public funding, and security regulations to re-establish control over infrastructure that once seemed on the verge of being lost. The most radical version involves corporations directly inheriting sovereignty from their balance sheets, ignoring the critical links beneath the surface and above the chips: energy, water, land, licenses, grid interconnection, and the law. It is a confrontation, not a conquest.

Advanced Control Available

By stacking the layers together, the pattern will appear.

It is essential to also explain its limitations, as most writers tend to cut corners on these.

At every level, the key factor influencing the value of safe assets under stress lies not only in credit but also in operational control—that is, the holder's ability to actually exercise their rights when circumstances deteriorate.

Controllability depends on the location of asset custody, the jurisdiction to which the issuer and registrar are governed, who can freeze or seize assets, whether assets can be transferred and settled if the clearing system is weaponized, whether sanctions render it inadmissible, and whether assets can be liquidated at the required scale and speed.

Government bonds held long in fragile custody chains, digital dollars that can be legally frozen, computing contracts subject to export licenses, and gold bars that are difficult to seize but difficult to move all lie at different points on this spectrum. Two assets with the same rating may have different levels of available control. That chasm is the frontier. Now, the boundaries are clearly defined. The risk of revocation has been priced in, and this has been confirmed. The International Monetary Fund has modeled geopolitical risks and sanctions. The European Central Bank and the European Committee on Systemic Risk closely monitor market fragmentation. Relevant literature has constructed news-based geopolitical risk factors and demonstrated that these factors can generate premiums. If this article merely conveys the message that political factors have been priced in, then it is merely borrowing from others' work and should be discarded. Open-ended questions, the only place where truly new things can exist, are a narrower and more demanding field. Most existing metrics are passive, built upon news, existing linkages, and attention indices. The question I cannot, and do not intend to, answer is: can a structured score—based on factors such as the asset's fixed characteristics, custody, jurisdiction, issuer's freezing rights, dependence on suppliers, finality of settlement, and admissibility under sanctions—and possess pre-emptive control, more accurately predict the asset's performance in actual withdrawal events, building upon existing geopolitical risk beta coefficients? If the structured score provides no additional information beyond known factors, then the available control variables are merely a reference, not a source of returns. It remains useful, but not in terms of excess returns. If it can add real incremental power to a series of real events (such as the 2022 freeze, successive sanctions designations, and chip control), and properly eliminate reactive factors, then there exists a security dimension that allocators have not yet measured. I don't know which one. And this is the most important sentence here. To answer this question, a cross-sectional event study of asset-level returns, spread changes, capital flows, discounts, and settlement outcomes is needed, eliminating known factors. Research data must come from institutional terminals, and the scoring framework must be constructed and validated asset by asset. This cannot be summarized in a single text, but is a research plan, and cannot be answered through reasoning or reading, both of which we have tried. To be sure, published works available externally do not seem to construct that precise pre-structured scoring. The ground appears open. The key question is whether it's open because it's valuable and undiscovered, or because it collapses into existing factors after accessing data. Rather than treating a conjecture as a discovery, it's better to present a clean sample directly. What can prove this wrong? Arguments that don't stand up to scrutiny are not analysis, but theology. Therefore, the following things will both kill and nourish it. If the share of dollar reserves remains stable, if Treasury conveniences stabilize or recover, if official demand proves resilient after proper calculation of offshore custody channels, if stablecoin demand remains too small or too volatile to be relevant even at the paper currency level, and if AI capabilities expand without significant bottlenecks in electricity, permissions, connectivity, or sovereignty, then this framework is flawed, or rather, exaggerated. If these assumptions hold true, then the so-called “decoupling” is actually more like a common combination of fiscal supply, regulation, and geopolitics. This conclusion deserves serious consideration because the strongest counterarguments are based precisely on relevant research from the IMF and the ECB. Geopolitical risks may already be fully priced in through known channels, so there may be no need to introduce new factors. If official sectors continue to withdraw from the long end, while private and stablecoin-related demand supports the front end; if gold accumulation continues to exceed what valuations can explain; if long-term convenience remains negative; if payment and computational sovereignty strengthens; and if Treasury bills and long-term bonds increasingly behave like two separate assets. These signals are specific, public, and dated, which is crucial because a framework you can't monitor is a framework you can't trade. Pay attention to the Treasury Department's monthly International Capital Report to understand the allocation of official and private funds. Pay attention to the Federal Reserve's weekly custody data for official Treasury balances. Pay attention to the IMF and ECB's quarterly reserve composition data (adjusted for valuation) of the dollar, euro, US Treasury bonds, and gold. Pay attention to stablecoin assets and their impact on Treasury yields. Closely monitor electricity, licensing, connectivity, and export policies, as these will determine whether infrastructure construction can proceed. Most importantly, pay attention to one thing. Under Warsh's leadership, this issue is no longer just theoretical. Warsh, sworn in in May, demonstrated both policy continuity and change, initiating discussions about streamlining the balance sheet, although debates continue about the extent to which the Fed can reduce its balance sheet. The ultimate guarantee of Treasury availability is not the bonds themselves, but the central bank's willingness to provide dollar loans to Treasury bonds through swap lines, FIMA repurchase mechanisms, and a broader liquidity architecture when the system needs funds most. The Fed is reassessing its size, and its crisis liquidity stance will be closely watched globally. If the Federal Reserve signals that it is unwilling to readily lend dollars during a crisis, the availability of the entire dollar reserve tier will change. This is the catalyst calendar. The key line: Sensational claims that a particular company is safer than the United States should be thoroughly refuted, as rejecting such claims is the very foundation of the argument. Such claims are often based on comparisons of credit default swap spreads, which fail when microstructures are involved. The U.S. single-guarantee market is small, and Federal Reserve research suggests that the spreads published in recent years may have seen little actual trading, making them unreliable as a measure of expectations. A deeper, more fundamental problem lies in the "cheapest delivery" option. In the 2023 debt ceiling crisis, suppose the cheapest deliverable bond in the default auction was a heavily discounted 30-year U.S. Treasury note, trading near the mid-$50 mark. Therefore, the amount of collateralized compensation (and the spread) reflects that discounted bond and the settlement mechanism, not any judgment on U.S. solvency. The net amount of outstanding collateralized guarantees against deliverable Treasury notes was negligible. Comparing this to corporate financial statements is a misclassification. Building a grand argument based on this would allow a fixed-income expert to disregard everything else. Sovereigns have not been replaced by corporations. What is happening now is much quieter and on a larger scale. The market is not replacing sovereign states with corporations, but rather repricing which tools, institutions, and channels control safeguards that were previously concentrated at the same level of security. The US dollar still dominates the financing arena. Treasury bonds still dominate the front-end market, now further strengthened by the digital dollar designed within a regulatory framework. Due to increased supply and declining convenience yields, long-term Treasury bonds are gradually moving away from their previous safe-haven role. Gold serves as a safety net, acting as a hedge against asset confiscation, but is difficult to mobilize on a large scale. Payments and computing are focal areas of competition for control among nations. Risk-free assets have not disappeared. This is decoupling. The job of asset allocators in this century is to purposefully seek each layer of security, rather than hoping that a single tool can provide all the protection. The premium the world paid for the old system has disappeared because the system itself is crumbling. The task now is to price these assets. The question I will explore next is: is available control merely the right way to view this task, or a way to profit from it? I intend to answer this question with data, not beliefs.