The next banking crisis may not begin with bad loans. It may begin with a smartphone notification. For generations, bank runs followed a relatively predictable pattern. Depositors heard rumours, queued outside branches and withdrew cash over several days or even weeks. Regulators had time to react. Central banks had time to organise rescue operations. Governments had time to calm markets. That world no longer exists. Silicon Valley Bank demonstrated something profoundly unsettling in March 2023. A quarter of its deposits disappeared in a single day. Had authorities not intervened immediately, withdrawals would likely have accelerated further the next morning. The speed of the collapse shocked everyone. The next one could be even faster.

Artificial intelligence, digital banking, social media and instant payments are transforming the very nature of liquidity risk. Depositors no longer need to drive to a branch. They no longer need to call their banker. Soon, they may not even need to make a decision themselves. Algorithms could automatically move deposits at the first sign of financial stress, negative headlines or a better interest rate elsewhere. Bank runs are becoming automated. And that changes everything. The danger becomes particularly acute in the current environment.

For months, global bond markets have been under pressure. Sovereign yields continue rising across the United States, Europe and Japan as governments issue record amounts of debt while inflation remains stubbornly elevated. Bond prices, by definition, move in the opposite direction. When yields rise, bond portfolios lose value. And banks remain among the largest holders of those portfolios. This is precisely what destroyed Silicon Valley Bank. The institution had invested heavily in long-duration US Treasuries and mortgage-backed securities during the era of near-zero interest rates. When the Federal Reserve began raising rates aggressively, the market value of those assets collapsed. On paper, the losses were manageable. In practice, they became fatal. Because deposits started leaving. To meet withdrawals, the bank was forced to sell assets that had lost substantial value. Unrealised losses suddenly became realised losses. Confidence disappeared. Depositors fled. The bank collapsed. The lesson was simple. A bank can survive asset losses. A bank can survive deposit withdrawals. What it often cannot survive is both at the same time. Today, that risk is re-emerging.

Bond markets are once again under pressure. Treasury yields continue moving higher. The market is increasingly questioning whether inflation will remain structurally elevated. Governments are borrowing at record levels. Central banks are no longer buyers of last resort. The result is a sustained decline in bond valuations. For banks, this creates a dangerous vulnerability. Many institutions continue holding large portfolios of sovereign bonds purchased when yields were significantly lower. As rates rise, the market value of those holdings declines. Under normal circumstances, banks can simply hold the securities until maturity and recover their principal. The problem arises when liquidity disappears. And liquidity can disappear remarkably quickly.

A negative rumour spreads on social media. Depositors begin questioning a bank’s financial health. A few large clients transfer funds elsewhere. Algorithms detect abnormal flows. More withdrawals follow. What once took weeks can now happen in hours. The bank suddenly needs cash. To obtain cash, it must sell assets. But the assets it owns are worth less than when they were purchased. The vicious circle begins. Asset sales generate losses. Losses damage confidence. Lower confidence accelerates withdrawals. More withdrawals require more asset sales. This is exactly the dynamic that transformed Silicon Valley Bank from a manageable problem into a systemic event.

And there is another layer of risk that markets may be underestimating. The sovereign debt market itself. For decades, government bonds were considered the safest assets in the financial system. They formed the foundation of bank liquidity management, collateral frameworks and regulatory capital requirements. That assumption is becoming less comfortable. If sovereign bond prices continue to fall due to massive issuance, inflation concerns, and deteriorating fiscal positions, banks face a double challenge: weaker asset values and more fragile liquidity buffers. In effect, the asset traditionally used to protect the banking system becomes a source of instability. This is why the current bond-market sell-off deserves far more attention than it receives. The issue is not merely higher yields. The issue is how higher yields affect balance sheets. And the implications extend beyond individual institutions. If several banks simultaneously need liquidity and begin selling long-duration government bonds, the pressure on sovereign markets could intensify dramatically. Falling bond prices would create additional losses, forcing further asset sales. The feedback loop could become self-reinforcing.

In an extreme scenario, this could even affect confidence in the dollar itself. Foreign investors and reserve managers hold trillions of dollars in US Treasuries because they are perceived as liquid and safe. If the market begins experiencing disorderly liquidation events and repeated banking stress linked to Treasury holdings, some investors may start questioning whether such concentration remains desirable. The risk is not a sudden collapse of the dollar. The risk is a gradual erosion of confidence. History shows that confidence rarely disappears overnight. It weakens progressively until a trigger event exposes vulnerabilities that everyone knew existed but preferred to ignore. Silicon Valley Bank was supposed to be a warning. The problem is that financial crises rarely repeat themselves exactly. They simply exploit the same weaknesses through new technologies. And in a world where billions of dollars can move with a single click, the next bank run may already be faster than the system designed to stop it.