The Caladan Weekly | The $2B Question: Why Crypto Buybacks Aren’t Working, and What Would

Jupiter’s @sssionggg floated a question in early 2026 that got many people talking: “What do you all think if we stop the JUP buyback? We spent more than $70M on buyback last year and the price obviously didn’t move much. We can use the $70M to give out for growth incentives for existing and new users. Should we do it?”

The responses split immediately. Some called buybacks a waste of treasury. Others called abandoning them a capitulation to short-termism.

But what made the conversation interesting is that it exposed something most people were feeling but not saying: the mechanism that was supposed to make utility tokens investable has NOT actually done that for most projects.

@layerggofficial compiled a dataset of 109 projects that have active, planned, or recently completed buyback programs. The numbers cover cumulative buyback percentages, buyback rates as a share of revenue, annualized buyback values, FDVs, and revenue figures. I dug into it to see what the data actually tells us, because the debate so far (from what I observed) has been mostly vibes.

The aggregate picture from this dataset: $2.1 billion in annualized buyback activity across $244.6 billion in combined FDV. That works out to an aggregate buyback yield of 0.86%.

For context, the S&P 500 averages roughly 2-3% in buyback yield. Crypto, with its dramatically higher volatility and structural supply pressures, is doing about a third of what equities do.

And the real story is worse than even that headline number suggests.

Almost all of the buyback activity is concentrated in a handful of projects.

5 projects account for 73.3% of all buyback dollar volume.

The top 10 capture 89.1%.

The top 18 account for 96.8%.

The remaining 91 projects split $67.7 million between them. Average annual buyback per project outside the top 18: $740,000. On tokens with FDVs often in the hundreds of millions. That’s not a buyback program. That’s a rounding error.

68 projects buy back less than $1 million per year. 56, more than half the dataset, have effectively zero realized buyback value despite having announced or planned programs.

The rate distribution tells the same story in a different way.

Layergg’s data includes a “1Y Buyback Rate (Revenue-Based Projection)” that takes current monthly revenue, annualizes it, multiplies by the project’s stated buyback allocation percentage, and divides by FDV. This gives you the forward-looking buyback yield if current revenue holds steady.

For two-thirds of these projects, even if they execute their buyback programs perfectly, the annualized impact on token supply is so small relative to FDV that it has no measurable price effect.

The revenue allocation problem is where it gets really interesting.

Some projects generate substantial revenue and barely direct any of it to buybacks. Others generate less but funnel nearly everything back.

A few standout comparisons show just how wide the gap is:

Pump.fun and AAVE both generate nine-figure annualized revenue. The buyback yield differential between them is roughly 100x. AAVE recently passed a governance proposal to begin buybacks, so this will likely shift, but the contrast as of January 2026 is stark. Lighter pulls in $91M and Arbitrum $84M in annualized revenue with zero flowing to buybacks.

The point isn’t that every project should route 97% of revenue to buybacks like Pump.fun. Projects need to fund development, hire people, grow. But the gap between what’s possible and what’s being executed is enormous.

Many of these teams are sitting on serious revenue and returning almost none of it to tokenholders. That’s not a buyback program. That’s a mere press release.

The “top-blast” problem might be the most important finding in this dataset.

Layergg tracks two different annualized buyback projections. The “Revenue-Based” projection uses current monthly revenue extrapolated forward. The “Time-Based” projection uses actual historical buyback spending.

When the time-based number is significantly higher than the revenue-based number, it means the project bought back much more aggressively in the past than current revenue can sustain.

In other words, the project spent its buyback budget during periods of higher revenue, which almost always correlates with higher token prices.

Jupiter is the case worth zooming in on, given it started this whole conversation. Its average monthly revenue over the past year was roughly $22.7M. Current monthly revenue has dropped to $9.98M.

At the current run-rate with a 50% buyback allocation, Jupiter’s forward annual buyback is roughly $59.9M, not the $136.4M it spent historically.

The 2.3x ratio means Jupiter bought the most tokens when they were the most expensive, which means fewer tokens were actually removed from circulation per dollar spent.

JUST is the extreme case: a 78.8x ratio. Revenue has essentially evaporated since the buyback was executed.

Now compare those to the steady executors, projects whose time-based and revenue-based rates are within 10% of each other:

These projects are buying at a roughly constant rate relative to their revenue, which means they’re not concentrating spending at cyclical peaks.

By definition, a consistent dollar-cost-averaging approach purchases more tokens when prices are low and fewer when prices are high. A project that spent $136M evenly throughout a volatile year removed far more tokens from circulation than one that spent $136M in a burst during the top.

The dollar value looks the same on a dashboard. The actual supply impact is wildly different.

Cumulative supply impact separates the serious from the performative.

Beyond annualized rates, the cumulative buyback percentage (total supply permanently removed) gives you a longer-term view of which projects have actually made a dent.

Only 9 projects have removed more than 10% of their total supply through buybacks. Several of them, notably Rollbit, Aerodrome, and BONK, have shown relative chart strength compared to the broader utility token space.

That’s not conclusive proof that buybacks caused the outperformance. But it does suggest that when cumulative supply removal crosses a certain threshold, something starts to shift in the supply/demand dynamics.

BNB is interesting here: 32.13% cumulative buyback on a $123B FDV, but a current 1Y buyback rate of 0%. Its burns are periodic and event-driven rather than continuous, and it generates $641M in annualized revenue that currently isn’t being directed to buybacks.

If BNB reactivated a continuous buyback program at even modest allocation rates, it would dwarf every other project in this dataset.

FDV size creates its own gravity.

Smaller projects have lower FDVs, so even modest revenue translates to meaningful buyback rates. But it also means the projects where buybacks “work” in percentage terms are often the ones where the absolute dollar volume is too small to attract attention.

Meanwhile, the large-cap tokens where capital allocators would most benefit from buyback signals generate rates too low to matter.

The $100M-$1B bracket appears to be the sweet spot: large enough to generate meaningful revenue, small enough that buybacks can actually move the needle on supply.

The structural backdrop makes all of this harder.

Before even getting to execution quality, there’s a market-level problem. Privacy coins, PoW tokens, and memecoins have dominated upside this cycle.

Most utility tokens are severely underperforming. In several recent crypto M&A transactions, tokens were discarded entirely while only equity was merged, reinforcing the narrative that tokens are disposable wrappers.

If Nasdaq moves to 24/5 trading as proposed, the gap between crypto and equities narrows further.

This creates a compounding challenge for buyback programs:

• Token unlocks and ongoing emissions dwarf buyback volume for most projects. If a team is buying back 2% of FDV annually while unlocking 10-15% through vesting and incentives, the buyback is just slowing the bleed, not reversing it.
• Memecoins, with fair launches and no vesting schedules, sidestep this entirely. A memecoin buyer doesn’t have to worry about a cliff unlock in three months dumping 8% of supply on the market. That structural simplicity is a genuine competitive advantage over utility tokens running buyback programs that can’t offset their own dilution.
• The “token is useless” narrative feeds on itself. As more M&A deals discard tokens in favor of equity, the perceived value of a buyback program shrinks, which makes tokens less attractive, which reinforces the narrative.

For buybacks to work as intended, they need to overcome all of this simultaneously. The data suggests very few projects currently do.

So what might actually work?

Based on this analysis, abandoning buybacks entirely seems like the wrong conclusion.

Given the options available to teams (buyback, dividend, or reinvestment), buybacks are likely still the most token-native strategy with the least regulatory risk. A token with buybacks versus one without, all else equal, is not a difficult choice.

But the execution probably needs to change fundamentally.

The flat “buy back X% of revenue every month regardless of price” approach has been tried. The data shows it leads to top-blasting: spending the most capital at the worst prices, creating an illusion of price support during euphoric periods, and then having insufficient firepower during drawdowns when the buying would actually matter.

One proposal that’s been circulating and seems worth serious consideration: calibrate buyback intensity to an on-chain P/E ratio.

Take an exponential moving average of revenue, annualize it, and adjust the buyback percentage based on where the token’s implied valuation sits:

• P/E under 4x: buy back 100% of revenue
• P/E of 4-6x: buy back 75%
• P/E of 6-8x: buy back 50%
• P/E of 8-10x: buy back 25%
• P/E above 10x: pause buybacks, accumulate a reserve

That reserve then deploys on dips, buying aggressively when the token trades well below its 90-day price EMA.

This approach would prevent top-blasting by automatically reducing buyback intensity when the token is richly valued, concentrate purchasing power at depressed prices where each dollar removes more supply, and build a war chest during euphoric periods that can be deployed as countercyclical support.

Whether this specific formula is the right one is debatable. But the broader principle seems sound based on what the data reveals: the steady executors (Meteora, BONK, Pump.fun, Rollbit) are already doing something like this intuitively, buying at consistent rates that naturally dollar-cost-average through cycles. Making it programmatic and price-aware would be the next step.

Where this leaves things.

Buybacks are getting a lot of heat right now. The frustration makes sense when you look at the numbers.

But the data points more toward an execution problem than a conceptual one. Most projects are running buyback programs that are either too small to matter (68 out of 109 buying back less than $1M/year), too poorly timed to be efficient (the top-blasters with 2-5x time/revenue divergence), or both.

The roughly 15 projects in this dataset that take buybacks seriously, that allocate meaningful revenue percentages, execute consistently rather than cyclically, and have removed real chunks of supply over time, seem to be getting results. The other 94 are using it as a marketing line.

The mechanism works. The implementation, for most of the market, doesn’t. Yet.