San Francisco Stablecoin Weekly Insights: The XYZ Coordinate System of 2026

Author: Charlie, Venture Partner @ Generative Ventures

Last week was Stablecoin Week in San Francisco, gathering leaders from the stablecoin industry around the world.

After going around, I have an increasingly strong feeling: everyone talks about "stablecoins," but they are actually discussing three different things.

In some venues, people talked about Circle's stock price, financial reports, and valuation reassessments.

In other venues, discussions revolved around agents, wallets, authorizations, payment protocols, and whether "AI really needs a card."

In some other venues, the discussions were less glamorous, focusing more on Brazil, Europe, corporate treasury, local inflows and outflows, non-USD liquidity, and the real issues in cross-border capital flows: each jurisdiction has its own compliance logic, banking restrictions, and settlement bottlenecks; transferring money is never as simple as "sending it over."

On the surface, everyone is talking about stablecoins, but they are actually discussing three different businesses.

This is also my biggest takeaway from this Stablecoin Week in San Francisco: by 2026, stablecoins will no longer be a single narrative but will gradually grow into a three-dimensional coordinate system.

The X-axis is Agentic Commerce.
The Y-axis is RWA and Credit.
The Z-axis is On-Chain FX.

Stablecoins are still a form of base currency, a common monetary substrate.

But the businesses that have been built around them have clearly diverged.

Circle's recent rise has made this change clearer.

In terms of financial reports, its data is certainly solid: by the end of 2025, the circulation of USDC will reach $75.3 billion, a year-on-year increase of 72%; in Q4 2025, the on-chain transaction volume of USDC will be $11.9 trillion, a year-on-year increase of 247%.

But what’s more noteworthy is that it is consciously transforming itself from a "stablecoin issuer" into a larger internet financial infrastructure: regulated stablecoins, tokenized currency funds, developer tools, Arc, Circle Payments Network—these pieces together are no longer just one coin, but an entire stack.

The old views are becoming insufficient. The stablecoin is still that asset, but the businesses that have grown around it are no longer the same.

1. X-axis: Agentic Commerce, the real change is not in payments, but in "who is spending"

I started writing about agentic commerce about a year ago. At that time, this term wasn't as popular, and many people's first reaction to the concept was still "AI helps you buy things" or "AI helps you browse Taobao."

But I have always felt that what’s truly worth watching is not this.

The most important change in agentic commerce is not that the shopping experience becomes smarter, but that a new type of actor has emerged in the commercial system: software with delegated intent.

This statement sounds a bit abstract, but the consequences it brings are very concrete.

In the past, e-commerce was centered around checkout. Whoever made the funnel shorter and the payment experience smoother was more likely to win.

But as agentic commerce progresses, the question is no longer just "how to pay," but "who can pay."

Who authorizes the agent? What are the boundaries of authorization? How much can it spend, and in what scenarios?

How is identity bound? How is risk control managed? How are disputes resolved? How is auditing documented?

If these questions are not solved, the payment track itself is not that important.

So I have been telling many friends in recent months that Circle's deepest moat is not just reserve income, not just distribution, not just regulatory dividends, but whether it can, in a deeper sense, become x402-native.

What I’m referring to is not "owning x402," because x402 is not something Circle invented.

I am talking about something else: if the future internet really grows a layer of machine-native payment layer, can Circle become the default dollar, the default wallet, the default settlement asset in that payment world?

This is a very critical distinction.

Because if you look at agentic commerce from today’s perspective, despite OpenClaw breaking boundaries, it largely still remains "AI placing orders for people."

But what will truly explode first may not be the most visible front-end scenario.

In many mature markets, agents will still prioritize using existing card organizations, bank transfers, or merchant-side vouchers, because these tracks are cheaper, more familiar, and easier to handle disputes.

What stablecoins will really first capture is likely a deeper layer:

Settlements between machines, pay-per-use APIs, pay-per-call for content and data, low-value high-frequency payments, autonomous treasury actions, and software-native global capital flows.

In other words, what stablecoins will win in agentic commerce first may not be "AI helping you buy a cup of coffee," but rather "software can finally spend like software."

This is also why I have repeatedly emphasized one statement:

If the internet will grow a layer of machine economy, the strategic high ground has never been about "issuing a dollar coin," but about becoming the wallet and settlement system that best aligns with machine behavior.

Not an extension of human checkout, but the starting point of software-native money.

2. Y-axis: RWA is still "old finance on-chain," but what’s most worth watching are the few truly on-chain native new assets

Today, the main body of RWA is still familiar financial assets moving on-chain.

The most typical example is U.S. Treasuries. Besides that, there are credits, commodities, funds, and increasingly more attempts at tokenized equities.

Ultimately, most RWA today is about switching existing financial products to a new track: making them more programmable, easier to distribute globally, and more efficient in settlement.

So I do not believe that the InfraFi proposed by Messari this year is the main line of current RWA.

But precisely because of this, I pay more attention to it.

Because in the entire RWA landscape, InfraFi may be one of the few things that truly feels like it has organically emerged from the on-chain world, rather than just a traditional financial product repackaged for redistribution.

A tokenized Treasury is essentially still a Treasury.

A tokenized stock is essentially still a stock.

They are certainly important and will grow larger, but their economic identity has not changed.

InfraFi is different.

It points to a class of assets or cash flows that have historically been difficult to standardize, continuously verify, and effectively finance, because on-chain verification, programmable ownership, and continuous data streams create opportunities to become truly investable objects.

This is why I place more importance on it.

Not because it is the largest today, but precisely because it is still small.

But it may represent a new way of generating assets.

In this direction, Arkreen’s EnergyFi is one of the cases I am currently most interested in.

My interest in it does not lie in the superficial narrative of "energy + RWA," but in how it may demonstrate how a truly on-chain native asset class should grow.

In the past, many infrastructure-type cash flows were not without value, but were too fragmented, too scattered, too reliant on offline verification, and too dependent on post-event aggregation, making it difficult to become an object that can be monitored frequently, continuously priced, and effectively financed. Often, it is not that the asset itself is lacking, but that the factual basis of the asset is too weak.

EnergyFi attempts to address precisely this issue.

If the underlying processes of energy production, usage, and settlement can continuously generate credible data;

If this data is not relayed through monthly reports, audit summaries, or third-party interpretations, but instead becomes a verifiable, callable, and traceable factual stream;

Then what is being financed is not just a packaged yield right, but a cash flow system that can be continuously verified.

Why is this important?

Because it addresses not just energy, not just DePIN, but the core pain points of private credit and even broader private investing.

The recent incident with Blue Owl Capital exposed many issues in private credit, which on the surface appear to be structural design issues, liquidity issues, and valuation issues, but fundamentally boil down to a very simple question: is the underlying information you receive real? Is it timely? Is it complete? Can it be continuously verified?

If not, then it is essentially garbage in, garbage out.

If the underlying facts are vague, lagging, and filtered, then no matter how sophisticated the upper structure is, it is merely repackaging opacity.

The reason why things like EnergyFi are worth serious consideration is not because they have a new concept, but because they may provide another option: not making "reports" prettier, but turning the underlying operational facts themselves into continuously verifiable data objects.

This will directly change underwriting, monitoring, and even the due diligence logic in primary and secondary markets.

If this path is successful, the significance of on-chain finance will not just be "adding a distribution channel," but adding a new truth layer.

In this sense, I see EnergyFi as one of the potential foundations for future private investing, rather than a marginal narrative.

Moreover, this line has a very realistic backdrop: the energy bottleneck of AI.

In the past few years, energy has often been treated as a backdrop to technological narratives.

But as AI enters the infrastructure competition phase, energy is no longer just a supporting condition but is becoming one of the hardest constraints.

If AI remains one of the most important industrial themes in the next decade, then systems surrounding energy production, financing, verification, and cash flow securitization will inevitably move from the margins to the center.

From this perspective, EnergyFi is not just adding a "green story" to RWA, but is more like a preview: can on-chain finance really cut into the skeletal parts of next-generation infrastructure financing?

3. Z-axis: On-Chain FX is not a payment issue, but a market structure issue

During this trip to San Francisco, I felt that one type of conversation that was most underestimated actually came from on-chain FX.

Perhaps it is because this group of people is closer to real liquidity, real corridors, and real balance sheets, so they generally speak more restrainedly and are less likely to frame things as a simple story of "faster payments" or "cheaper transfers."

My biggest impression is that many people's understanding of on-chain FX still remains at the extension of cross-border payments.

But the real difficulty in this matter has never been about "sending money faster"; the real difficulty lies in market structure.

Foreign exchange has never been just an information transmission issue, nor just a settlement issue.

It is primarily a balance sheet, funding, and liquidity issue.

Blockchain excels at atomic settlement; mature FX markets excel at netting.

The former is clean, direct, and minimizes trust, but is very capital-intensive; the latter is complex in its system but saves balance sheets significantly.

If every transaction must be gross settled, capital will be locked up.

The efficiency of market makers will drop, spreads will widen, and depth will fail to develop.

So this week, I repeatedly heard a very clear judgment in multiple discussions about on-chain FX: what is currently truly blocking this market is not that contracts cannot be written, not that wallets are inadequate, and not just compliance issues, but capital efficiency.

This judgment is important because it will directly change how you view this track.

If you only understand on-chain FX as "on-chain currency exchange," it is easy to draw an overly simplistic conclusion: do more infrastructure, issue more local stablecoins, and add more trading pairs, and the market will naturally come.

But reality is not like that at all.

A more suitable way to understand it is to break it down into three levels.

The first level is fiat-to-fiat: this is the largest market but also the hardest to immediately leverage, as traditional institutions have already served it deeply.

The second level is fiat-to-stable and stable-to-fiat: this is actually the part with the most obvious growth today, especially in scenarios like emerging market treasury, remittances, and corporate settlements.

The third level is stable-to-stable: it is currently not large in scale, but it may be the true endgame. Because only at this level does FX no longer become "using crypto to assist access to old tracks," but begins to resemble an internet-native foreign exchange market.

This layered approach also clarifies the boundaries of opportunity.

What will be transformed first, as I mentioned in Airwallex's founder on where stablecoins went wrong, will not be the deepest G10 interdealer market, but rather the long tail that the old system has consistently failed to serve: small and medium enterprises, exporters, cross-border platforms, freelancers, migrant corridors, and various participants without prime broker relationships or large balance sheets, but who indeed have ongoing cross-border needs.

I heard several representative examples this time.

One team mentioned that Starlink's treasury corridor in Latin America reduced the settlement cycle from several days or even weeks to about 35 minutes.

Another team plans to launch in mid-2025, achieving $1 billion in transaction volume in six months, and then another billion in the following three months, primarily serving emerging market FX and regulatory demands.

These cases illustrate one thing: the growth of on-chain FX does not emerge from the center but grows from the margins.

It does not start by replacing the deepest markets but first smooths out those areas that the old system has consistently failed to serve.

Interestingly, these discussions have also made me increasingly believe in another thing: on-chain FX is primarily not a supply-side infrastructure issue, but a demand aggregation issue.

I heard a statement at the meeting that left a deep impression on me: start with the chicken, not the egg.

The meaning is, do not get lost in designing a perfect liquidity market from the start; first capture the real flows.

If you can aggregate real demand along corridors through neobanks, PSPs, remittance platforms, treasury software, or products that already have distribution capabilities, market makers will naturally come.

Conversely, without stable flows, adding another trading pair or venue will also struggle to create real depth.

This perspective is particularly important.

Because it implies that the winners in on-chain FX may not come from traditional exchanges but more likely from the orchestration layer: they may not necessarily consume all liquidity themselves, but they will coordinate compliant inflows and outflows, aggregate demand, perform smart routing, and layer netting and credit at the right time.

Ultimately, once you acknowledge that the bottleneck of on-chain FX is capital efficiency, you are actually not far from credit.

4. Non-USD stablecoins are not just about issuing a coin; the real challenge is corridor capability

If I had to choose which line is most easily underestimated this week, I would pick non-USD stablecoins.

Because it is always easy to present on PPT: every country, every market, seems to need its own stablecoin.

Logically, this is certainly correct, but in reality, this is a business that is much harder than USD stablecoins.

Why are USD stablecoins strong? Because in many countries, the dollar is naturally a stronger store of value.

When issues like local currency volatility, capital controls, and banking capabilities overlap, the dollar automatically becomes the default option. So it is not surprising that today’s stablecoin world is centered around the dollar.

The problem with non-USD stablecoins is that many people think "issuing" is the product, but in fact, issuance is just the easiest step.

What is truly difficult is local banking relationships, offshore liquidity, market makers, same-name pay-in/pay-out, connecting with local payment networks like SEPA and PIX, navigating regulations corridor by corridor, and creating a user experience competitive enough to rival Wise, Revolut, local PSPs, and bank transfers.

More critically, you must answer a question that is much harder than "can we issue a coin": why would users want to hold local currency on-chain?

This question is actually very sharp.

Because in many markets, the demand for local currency is primarily not a store of value demand, but a payout demand.

Users may still want to use dollars as a value storage tool, but in payroll, invoices, supplier payments, taxes, and domestic spending, they must revert to local currency.

This means that what non-USD stablecoins really need to do is not to make the market "recognize this token," but to become a bridge: one end connected to offshore liquidity and the other end connected to local payment systems and real business flows.

Brazil is a very typical example.

What is truly valuable is not the fact that "BRL is on-chain," but whether it can bridge the onshore and offshore worlds in a market where cross-border transactions are already extremely painful, where large order books thin out, and where fiat cross-border holdings are highly restricted.

From this perspective, the token itself is not that important; what matters is the entire corridor architecture.

I also feel that the market generally underestimates the path dependency of non-USD stablecoins.
USDT and USDC did not grow because of logical consistency; they grew because they hit real liquidity and distribution flywheels.

Non-USD stablecoins must also find their own growth flywheel and catalyst.

If they cannot find that catalytic point, they will remain stuck in the stage of "theoretically correct but practically thin."

So I completely agree that non-USD stablecoins are very important, but they are by no means a natural extension of USD stablecoins.

They are a business that is slower, harder, and tests corridor operational capabilities more.

5. What is truly worth competing for in 2026 is not the coin itself, but the control points outside the coin

If we put these three lines together, the coordinate system for stablecoins in 2026 is already quite clear.

Agentic commerce is fundamentally an authorization issue.

RWA and on-chain credit face verification challenges.

On-chain FX ultimately revolves around capital efficiency.

Stablecoins are still a common base layer, but the real competition going forward is no longer just about "who issues more."

Some will win on intent and permission.

Some will win on truth and underwriting.

Some will win on corridor liquidity, routing, and netting.

This is why I increasingly feel that the term "stablecoin company" itself is becoming less precise.

The market has previously focused on issuance. But in the next phase, what really matters is who controls the layer of structure outside the stablecoin: authorization, verification, credit, liquidity.

In this sense, Circle's recent revaluation is certainly worth watching, but it should not be viewed merely as a stock story.

It is more like a signal: the capital markets are beginning to vaguely realize that stablecoins are no longer just a coin, but the monetary base for three different businesses.

Stablecoins are certainly still assets.

But the real business going forward is about who decides how this money is authorized, how it is verified, how it is financed, and how it is exchanged.