The question this cycle finally answers

As of the end of February 2026, Bitcoin has experienced a 50% peak-to-trough drawdown, while realised volatility remains lower than the comparable phase of 2018/2022. US spot Bitcoin ETFs have seen negative $3 billion net flows over the first 8 weeks of the year, and total digital asset ETP AuM stands at $133 billion. The signal is familiar: price shakes conviction. The interesting part is what happens next.

Every downturn in this market reveals the same uncomfortable truth: conviction is easy to declare and expensive to maintain. Since 2013, I have watched wave after wave of participants arrive with ambition and leave with excuses. The projects, the promises, the pitch decks. They follow the same arc. What changes from cycle to cycle is not the pattern, but the quality of the doubt.

What’s different this time is not mood or price action. It is participant composition and infrastructure utilisation. Previous bear markets were dominated by reactive scepticism: price falls, headlines follow, and the chorus of obituaries begins. Bitcoin drops and the narrative writes itself: the experiment is over, rational markets have spoken. That kind of criticism dissolves as quickly as it forms. Markets trade narratives; infrastructure compounds silently.

What I hear now is a better question and it deserves a better answer: after fifteen years, has crypto produced something genuinely useful? Is anything real being built? The honest answer comes in two parts.

First: Bitcoin works. It has done so through every stress test that matters: technical, market, political, and behavioural. When CoinShares entered this market in 2013, Bitcoin’s network value was closer to a mid-cap equity than a macro asset. Today it is routinely discussed in the same breath as sovereign reserves and system liquidity. Volatility is a feature of monetisation events, not evidence of technological failure.

Second: much of what occupied the last cycle did not produce anything durable: gaming tokens, speculative NFT markets, “yield” loops that were really just leverage dressed in new language. The industry confused financial engineering with utility and leverage with innovation. Markets eventually correct both. When everything is for sale and the decks keep arriving, it reflects something real about where we are in the cycle. The builders who matter keep building. The rest look for the exit. Nothing has really changed since Buffett's famous quote about the tide going out. 

Now to the part that makes the timing of today’s doubt almost perfectly wrong.

The agentic economy is increasingly no longer a concept. It is infrastructure being built in real time. We are moving from an internet of information to an internet of decisions. McKinsey — echoing projections from Musk and CZ —, projects AI agents could orchestrate trillions of dollars of commerce by 2030. In 2025, Visa, Mastercard, Google and Stripe each launched agentic payment frameworks within months of one another. Meanwhile, the global economy remains anchored in consumer demand. Roughly 70% of the US GDP is consumption and payments still carry meaningful friction in a world that is about to be defined by machine-to-machine execution. I am leaving the privacy argument aside for now, but it is certainly something we will need to consider.. 

This creates a new, very practical engineering problem: How do autonomous systems transact: cheaply, globally, and at machine speed?

That problem has a settlement-layer problem at its core.

Agents need a trustless settlement layer

Consider the arithmetic. A $0.20 card processing fee applied to a $0.003 micro-transaction between two AI agents is ~6,600% overhead. Traditional banking infrastructure was built on an assumption: a human on one end of every transaction meaning an identity, a passport, an account, an approval chain, and human-paced dispute resolution. That architecture does not scale to a world where each person may operate hundreds, eventually thousands, of autonomous agents acting on their behalf based on their reputation (for instance, new ERC8004 standard), negotiating and executing performance based transactions continuously supported by escrow mechanisms (like Coinbase’s X402 protocol).

Card networks solved trust between humans. The next economy requires trust between machines.

The primary financial infrastructure that handles this natively is programmable, on-chain money: instantaneous, auditable, composable, and able to operate without human approval at every step. This is not a speculative application of crypto. It is a structural requirement of the economy arriving whether incumbent payment rails are ready for it or not.

For institutional investors, that distinction matters. Digital assets are no longer only a directional bet on price; they are increasingly a form of  exposure to a set of rails: settlement, collateral mobility, and programmable transfer of value. Stablecoins are not competing with banks; they are competing with latency. The question is no longer whether crypto has a use case, it is which industries will struggle to operate without programmable settlement.

This is precisely where CoinShares is positioned.

CoinShares has operated through multiple cycles since 2013, not by trying to guess which narrative will outperform next week, but by building regulated access, institutional-grade product design, risk management, and liquidity expertise around the underlying rails. We do not need every token to succeed. We need the infrastructure to be used and to be investable, transparent, and properly governed for professional capital.

Durable infrastructure is rarely built during euphoria. It is built during doubt, by those who remain because they understand what they are working toward. What has not changed is the direction of travel: toward a financial system that is more programmable, more transparent, and more accessible. The agentic economy does not change that thesis. It accelerates it.

Previous cycles asked whether Bitcoin could survive. This cycle is beginning to answer a different question: whether the next economy can function efficiently without programmable settlement.