NEW YORK – When Strategy, once known as MicroStrategy, began issuing shares above the value of its bitcoin holdings to purchase even more cryptocurrency, the market greeted it as financial engineering at its most creative. Investors who bought at that peak now hold a position worth roughly 20 cents on the dollar. The company has fallen around 80 percent from its high, its premium to net asset value reduced to nearly zero.
The decline is not the story. The story is that it mirrors, almost exactly, what happened to investors who bought Sandisk (SNDK) at the top of its 4,000 percent year-over-year rally, and to funds that accumulated silver positions as the metal pushed past $120 an ounce in January before surrendering half those gains. Each instrument told a different story. Each told the same one underneath.
A CoinDesk analysis published Monday by James Van Straten laid out the case plainly: structural trends can be real while their valuations remain cyclical. The distinction is elementary on paper. In practice, it is the one investors consistently fail to maintain.
The AI semiconductor boom was unmistakably real. Amazon, Google, Microsoft, and Meta spent the equivalent of small national economies building out data centers requiring hundreds of thousands of AI accelerators, along with the high-bandwidth memory and NAND flash storage that run alongside them. Micron Technology (MU) gained roughly 700 percent year-over-year at its peak. Sandisk, which returned as an independent company with exposure to flash storage, did even better on a percentage basis before giving back most of those gains. The underlying demand did not evaporate. Hyperscaler capital expenditure schedules remain near record levels. What happened to chip stock valuations was not the cancellation of the AI buildout but the exhaustion of the premium the market was willing to pay for front-running it.
Precious metals followed the same contour. Silver accelerated in early 2026 on what traders called the debasement trade – the thesis that sustained government deficits and persistent inflation would erode fiat purchasing power faster than central banks would admit. The metal’s industrial applications in solar panels and electric vehicle components gave the structural argument additional support. By January, silver was above $120 an ounce. By spring, it had surrendered more than half those gains. The debasement thesis was not disproved by the decline. What changed was the pace of speculative accumulation relative to what the underlying thesis actually justified.
Strategy’s situation added a recursive wrinkle. The company developed what analysts described as an “infinite money glitch”: a loop in which the market valued its shares above the net asset value of the bitcoin it held, allowing it to issue new equity at that premium and use the proceeds to acquire more bitcoin, which temporarily justified the premium again. As long as bitcoin was rising and the premium held, each step looked rational. When the loop broke, it broke completely. Even after bitcoin fell below $60,000 and triggered a reassessment of speculative positions built around the cryptocurrency, Strategy’s losses exceeded those of the underlying asset it was designed to track.
The AI chip sector provides perhaps the clearest case study in separating trend from valuation. Data center operators were genuinely buying. The disagreement was never about whether demand existed, but about how long the growth trajectory would stay steep enough to justify prices that already reflected a decade of above-trend expansion compressed into twelve months. When the trajectory remained strong but the rate of acceleration began to slow, that was enough to reset stocks priced as if the acceleration would continue indefinitely.
What makes the current cycle distinctive is the concurrence of multiple asset classes displaying the same dynamic simultaneously – semiconductors, precious metals, and digital assets, each with its own structural thesis, each arriving at a version of the same overcrowding problem. Whether the causes are related – excess capital flowing into structural narratives simultaneously – or incidental – unrelated sectors peaking at the same time – is not yet clear. The gold correction earlier this year, which saw the metal fall more than 20 percent from a January peak above $5,000 before the structural bid began reasserting itself, added another case study to the same pattern.
The difficulty of trading the distinction between trend and valuation is not primarily intellectual. Selling a position that has returned 1,000 percent because you believe it is overvalued requires the conviction to be wrong for an unknown period before the conviction is vindicated. Most institutional investors cannot sustain that posture; their clients cannot either. The result is that capital accumulates into structural narratives at precisely the point where the gap between the narrative and what the current price implies has grown widest. That is the mechanism that transforms genuine paradigm shifts into speculative bubbles and back again.
What the CoinDesk analysis leaves unresolved – deliberately, it seems – is whether the AI infrastructure buildout, bitcoin, and precious metals are mid-cycle assets that have cleared an overcrowding problem and can resume their structural trajectories, or whether the most aggressive assumptions embedded in their peaks are already in the rearview mirror. Market participants disagree sharply, which is why none of these assets have recovered to their highs. The disagreement, rather than any specific data point, is the condition the market is in now.
