Bitcoin's Power Curve: Generational Bottom or Broken Model?

↓ Jump to the video and timestamps

I'll be honest with you. I let myself drift. I get caught up in supercycle narratives, I see the chirps on Twitter saying the power curve is broken, and part of me starts entertaining it.

That's exactly why I keep bringing Matthew Mežinskis back every quarter. He shows up with the data and grounds me before I do something stupid with my thinking.

This time around the naysayers are louder than usual. Bitcoin has been grinding sideways to down from the 2025 highs, and the "power law is pseudoscience" crowd has been feeling themselves. Matthew walked me through every model comparison on screen, live, in real time, and the conclusion is the same one we've reached before, just at an even more extreme reading: we are near the cheapest Bitcoin has ever been relative to its long-run power regression.

The last time we talked, in February, I called it a generational buying opportunity. Matthew told me then. Now he's telling me it's even lower.

The monetary base side of the conversation carries equal weight. The global base has been coiling in a historically tight band for four years, Powell deliberately left Warsh runway to monetize if things crack, and the AI buildout is being treated as an existential national priority that will get backstopped by the full force of the federal government. That collision point between Bitcoin's power curve and the exponential fiat machine is the most important open question in macro right now. We don't have the answer, but this episode gets as close as anyone has to framing it correctly.

Key takeaways

• Bitcoin is near the zeroth percentile of its historical power curve. By Matthew's regression, the trend price sits around $142,000. We're trading at roughly 0.446 times trend, under 50% of the long-run line, the cheapest reading relative to the power regression in years.
• The power curve has been stable for over a decade. A regression calculated on data through 2016 only, no new information for ten years, projects a price within about $15,000-$17,000 of today's model. No other model (stock-to-flow, exponential) comes within an order of magnitude of that consistency.
• The global monetary base is coiling after four years of tightening. Down from over $30 trillion at the COVID peak to roughly $26.2 trillion at the end of Q1, growing at historically slow rates since 2008. The TA read is suppressed volatility, a band getting ready to break.
• Powell left Warsh room to monetize. The Fed owned 20% of US federal debt at the pandemic peak. As of June 30th, Mežinskis puts that at 11.38%, with $4.5 trillion of the $39.46 trillion in federal debt on the Fed's books. That gap is runway.
• Bitcoin is gaining ground on Fedwire regardless of price. The ratio of Fedwire's annual settlement volume to Bitcoin's on-chain volume improved from 134x in November 2023 to roughly 47.6x today, even as dollar prices have moved sideways or lower from the 2025 highs.
• The civilizational question is still open. Bitcoin's ~40% CAGR will eventually intersect with the stock market's accelerating exponential growth. Whether Bitcoin pulls TradFi onto a stable power-law path or TradFi co-opts Bitcoin into the volatile exponential machine is the most important macro question of the next decade.

Where We Are on the Power Curve Right Now

The opening chart walk-through in this episode should put the Twitter skeptics to rest, at least for anyone willing to look at the data honestly.

Matthew pulls up the log-log regression that he's been running since 2018. On log-log, Bitcoin's price plots as a straight line. That's not an aesthetic choice, it's what the data shows, with an R-squared of 96.1% by his model. No other regression fit, not exponential, not stock-to-flow, not linear, not logarithmic, comes close to that consistency.

Where we are right now on that regression is stark. Trend price is approximately $142,000 at the time of recording, per Matthew's model. Bitcoin is trading at about 0.446 times that trend, under 50% of the long-run line. Matthew calls this the Q0, or zeroth percentile, the worst data point the probability range can paint.

We've been here before, briefly, during the COVID crash and again during the SBF/FTX collapse. Each of those instances punched a new lower bound and expanded the outer edge of probabilities. We've now done it again.

In February, I said generational buying opportunity. Matthew told me then. Now he's telling me it's even lower than it was in February. His framing: "statistically, we are in a very, very bullish phase of Bitcoin's growth relative to the power curve. I mean, you can't really get much cheaper."

Power Curve vs. Exponential, Why the Distinction Actually Matters

I played devil's advocate on this one, and I meant it sincerely, because I've heard these arguments and I think you have to take them seriously before you reject them.

The steelman version: you're drawing lines, you're adjusting the model, you're making it fit. Are you the next Plan B? Is this pseudoscience?

Matthew's answer is to just run the comparisons. Exponential model, right now, would put Bitcoin at roughly $600,000. Plan B's 2020 stock-to-flow model, per Matthew's chart, projects $6.58 million. Plan B's 2019 model comes in around $800,000. Bitcoin hasn't touched the exponential trend since mid-2022.

Stock-to-flow's stair-step model has been inaccurate since before the ETF approval in early 2024, and the misses are off by orders of magnitude.

The power curve, calculated using only data available through 2016, no new information for ten years, still projects a price within about $15,000-$17,000 of the current model. Matthew's reaction to this on screen was the same as mine: "It's literally the same number."

The conceptual point underneath all of this matters too. Exponential growth means a constant CAGR, like the rule of 72, doubling at the same interval forever. The stock market does that. Gold does it. Bonds do it. Bitcoin is not doing that right now. Bitcoin's growth is currently declining, slowly, which makes it a power curve: for every 13% increase in Bitcoin's age, the price doubles.

Right now that works out to roughly a doubling every two years, or about a 40% CAGR. That CAGR is declining over time, but it's declining predictably and it's still enormous compared to anything else in traditional markets.

If the power curve breaks, Matthew is clear on the direction: it breaks to the upside, not the downside. Breaking the power curve means going exponential, which means faster growth, not slower. The critics who say "the power law is broken" to mean Bitcoin is doomed have it exactly backwards.

The Global Monetary Base: $26 Trillion and Coiling

The worldwide monetary base dashboard Matthew runs is one of the most useful macro charts I've seen anywhere. Here's where we are.

The global base peaked at over $30 trillion during COVID. It's now at roughly $26.2 trillion at the end of Q1, per Matthew's aggregation of central bank data. That's slightly down even from Q4 2025.

The CAGR since 2008 comes in around 7.6% in dollar terms, noticeably slower than the 10-12% pace from 1971 to 2008. The ECB is at historically low quantiles on its balance sheet. The Fed is at $6.74 trillion as of last Wednesday at the time of recording.

The TA chartist in me looks at that tightening band and sees one thing: a coil. Four years of compressing within a narrowing range, growing at slow rates, with the printing press sitting right there if conditions deteriorate.

Matthew described it as trend historically low levels for central bank balance sheets. I called it suppressed volatility. We both meant the same thing: this band is getting ready to move.

The directional read is not complicated. We know central banks reach for the printing press when things crack. The press is loaded. The band is tight.

The Fed has been a reasonably disciplined steward of the balance sheet relative to the panic-printing of 2020-2021. That's not a forever state.

One nuance worth noting: Matthew also showed that the money base actually went down over the last four years while Bitcoin ran a bull cycle. That tells you something about where Bitcoin is in its adoption curve. It's still small enough that the power trend is a more important driver than the money base.

The money base matters at the macro overlay level. The power curve is still the primary signal for Bitcoin's direction.

What Powell Left Warsh

The 100-year chart of the Federal Reserve's ownership of US federal debt is one of Matthew's best slides.

At the start of World War II, gold was roughly 86-87% of the Fed's balance sheet. The Fed carried almost no government bonds. Post-Bretton Woods, post-Nixon, Treasury buying became the dominant position and ownership climbed. At the pandemic peak in 2021, the Fed owned approximately 20% of all US federal debt, the highest monetization ratio in modern history.

As of June 30th, by Matthew's calculation, that's down to 11.38%, with $4.5 trillion out of $39.46 trillion in federal debt on the Fed's books. That 8-9 percentage point decline is the runway Powell preserved. Warsh didn't inherit a depleted balance sheet. He inherited a balance sheet with meaningful room to expand if conditions force the issue.

The Fed also still carries its gold at the statutory value of $42 per ounce on the H.4.1, a detail Matthew flagged as worth keeping in mind when you look at what the actual mark-to-market picture would look like.

Matthew's framing: "Powell gave Warsh plenty of runway to work with should they need to monetize again." The chart simply shows room to expand if conditions require it.

The SLR reform angle Marty raised, pushing credit expansion to commercial banks rather than through the Fed's own balance sheet, is the mechanism to watch in coming quarters. The thesis is that the Fed doesn't want to be the visible actor in the next round of monetary expansion, so they're loosening the borrowing constraints on banks instead. Matthew flagged this as something he wants a better handle on before the next call.

Bitcoin vs. Fedwire: The Money Chart

This is the chart I keep coming back to, and Matthew agrees it's one of the best in the whole deck.

Fedwire is the base settlement layer of the US financial system. The Fed holds roughly $2-4 trillion in reserves to backstop it. That reserve base turns over at a velocity that produces approximately $1.2 quadrillion in annual settlement volume, per Matthew's figures.

That's the number Bitcoin is actually competing with, not Visa, not Mastercard. The base layer.

Bitcoin's trailing 12-month on-chain transfer volume record was set in 2022 at roughly $60 trillion. That was at $16,000-$17,000 Bitcoin, meaning it was volume, not price. Currently, Bitcoin is running at about $24 trillion a year.

The ratio of Fedwire volume to Bitcoin volume: 134x in November 2023. Now, 47.6x. Bitcoin's dollar price has moved sideways to lower from the 2025 highs. The ratio has gotten significantly better anyway. As Matthew put it: "Price is even down, right, from say 2025, and the ratio's still getting better."

Bitcoin is gaining ground on the base layer of the US financial system in real value transfer terms, independent of price moves. Matthew noted the TPS comparison as well; Bitcoin's layer-1 throughput is comparable to Fedwire's. All Bitcoin needs to do to close the gap is rise in price.

The throughput is already there. His read: "Pretty high signal right there." Mine too.

The Civilizational Stakes: Power Law vs. Exponential Endgame

This is the thread I keep steering back to in every conversation with Matthew, and it's the one that matters most.

Bitcoin's ~40% CAGR is currently higher than the stock market's 12-14% (with dividends). Top US tech stocks are running at 22-25% even before the AI acceleration. The AI boom is pulling that number higher.

At some point, these growth curves intersect. When they do, one of two things happens.

Scenario one: Bitcoin pulls TradFi onto a more stable power-law growth path. That's the cypherpunk world. Distributed networks, real collateral, no rehypothecation, people can actually withdraw their coins, multisig on-chain custody is the norm. Matthew described it as "Bitcoin, like a black hole, pulls the whole TradFi market into its really interesting, more stable growth rate."

Scenario two: TradFi co-opts Bitcoin. Exponential debt contracts pile up against fixed supply. Musical chairs on limited coins. Withdrawal shutdowns. ETF-only custody becomes standard. Bitcoin starts mirroring the stock market, long periods of nothing punctuated by explosive moves and violent corrections. More volatility, not less.

Matthew's honest position and mine: the jury is still out. He leans toward hoping for scenario one. So do I.

But the intersection point may be getting pulled forward faster than anyone expected, because AI capex is being treated as an existential national priority. The federal government will backstop it. The Fed, tied more closely to the Treasury under the current administration's intent, will be the instrument of that backstop. That accelerates the collision.

One thing I've been writing about in the newsletter that connects here: people, especially younger ones, look at this broken, K-shaped economy and conclude that capitalism failed them. I think they're identifying a real problem and misidentifying the cause.

We live in a quasi-socialist experiment, not a capitalist one, because the most important tool for coordinating economic activity, money, is centrally controlled and can be debased at will. The income inequality and the asset-price inflation that Gen Z is revolting against are fiat outputs, not capitalism outputs. Bitcoin is the fix to the system they correctly identify as broken.

I also shared the Henry Adams curve in this conversation, which tracks energy consumption per capita against a long-run historical trend. We've been underperforming that trend since roughly 1971. The stock market is going up exponentially. The monetary base similarly.

But the base layer of the real economy, the energy system, has been starved of capital. Matthew flagged the post-Fukushima nuclear exits in Germany and elsewhere as a major contributor to that last dip. You can have all the fiat paper gains in the world. If the energy base layer doesn't keep up, the whole thing is on borrowed time.

That neglect of the base layer has a hard limit. It's another reason the exponential doom loop runs into a wall somewhere, even if nobody can tell you exactly when.

On the cypherpunk question: I'm bullish, and I said so. The rumors of Bitcoin's cypherpunk roots being ripped out and dead are greatly exaggerated. I recorded with Francis from Bull Bitcoin a few weeks ago and came away jacked up.

ARCs are launching. Arc Labs is working on covenant solutions. Square is enabling payments for merchants this year. The suitcoiners are loud, but the builders are building. That matters.

Log-Periodic Cycles and the 2028 Price Targets

Matthew collaborated with a researcher he described as having founded a Bitcoin science institute in Torino, Italy, on log-periodic price cycle analysis. The concept: within the power curve's long-run trend, can you mathematically identify and measure the oscillating cycles? The answer, by this analysis, appears to be yes.

Running the log-periodic formula without constraints produced five cycles. The three dominant ones, ranked by amplitude, were at 2.11 years, 3.7 years, and 7.6 years. These emerged purely from fitting the math to Bitcoin's price history in log space. No predetermined cycle length was assumed.

I want to be careful with how I present what comes next, because Matthew was careful too. These are model outputs with explicit probability framing. They are not predictions. The log-periodic model is more volatile than the pure power curve, and Matthew made that caveat clearly on tape.

With that scaffolding in place: by this model, end-of-2028 targets look like this. The pure power curve, if Bitcoin is on trend, projects approximately $300,000. The log-periodic model's central projection is $520,000.

The Q90 reading, meaning 90% of prior observations would fall below this level, comes in at $778,000. The Q100, which corresponds to an amplitude last seen in the 2013 cycle, and which Matthew described as "extreme upside" that he'd be "very, very cautious" about, is $2.2 million.

Matthew's framing on the Q100 was honest: we haven't been at that oscillation amplitude since 2013. Treat it accordingly. His framing on the power curve $300,000: if we're on trend at end of 2028, that's where the line sits.

The log-periodic work is directional research, not a clock. I find it interesting. I don't trade off it. Neither does Matthew.

About Matthew Mežinskis

Matthew Mežinskis is an independent macro researcher and host who has tracked Bitcoin's price regression and global monetary base data since 2018. He produces regular streams and data updates covering the power curve, central bank balance sheets, and on-chain value transfer metrics. He is a recurring guest on TFTC and collaborates with researchers on log-periodic cycle analysis and network adoption models. He is based in Europe.

Sources mentioned

• Fedwire Services annual statistics (Federal Reserve): source for Fedwire annual settlement volume figures
• Federal Reserve H.4.1 Statistical Release: Fed balance sheet total, composition, and gold valuation at $42/oz
• Federal Reserve Bank Term Funding Program (BTFP): the post-2023 banking crisis swap facility Marty references
• FRED: Federal Debt Held by Federal Reserve Banks: underlying data for the Fed ownership of US federal debt percentage
• Put America Back on the Henry Adams Curve (TFTC): the energy per capita framework Marty shares in the episode
• Fedwire Shows What Bitcoin Is Actually Competing With (TFTC): prior TFTC coverage of the Fedwire comparison
• Stargate's 7-Gigawatt AI Buildout Is Repricing Power Beneath Bitcoin Miners (TFTC): context on AI capex pulling the exponential collision point forward
• Nvidia as Guarantor: OpenAI's Ohio Bet Exposes AI Capex's Debt Problem (TFTC): the federal backstop dynamic for AI infrastructure
• AI Earnings Boom Hides $549B Depreciation Bill Coming Due (TFTC): the super-exponential stock market growth Matthew describes

Watch the conversation

Timestamps

• 0:07 - Why Bitcoin wins in a fiat debasement world
• 2:46 - Power curve update: even lower than February
• 7:32 - Walking the regression bands live
• 18:48 - Power curve vs. exponential and stock-to-flow head to head
• 28:21 - Sponsors: Unchained
• 39:38 - The exponential intersection point and AI acceleration
• 47:59 - Sponsor: Aven
• 48:53 - Sponsor: CrowdHealth
• 56:37 - Cypherpunk roots, ARCs, and Bitcoin payments
• 1:06:06 - Suitcoiners, Europe, and the future of financial engineering
• 1:10:29 - Gen Z, fiat capitalism, and the DSA problem
• 1:15:34 - Global monetary base: $26T and coiling
• 1:20:19 - Henry Adams curve and energy underinvestment
• 1:28:39 - BTFP, Fed balance sheet composition, and Powell's runway for Warsh
• 1:32:00 - Bitcoin vs. Fedwire: the money chart
• 1:34:49 - The 134x to 47.6x ratio improvement
• 1:40:54 - Log-periodic cycles and the 2028 bull juice

Sponsors

• Unchained: the leader in collaborative multisig custody, securing over $12 billion in Bitcoin for more than 12,000 clients. You hold 2 keys, they hold 1. Trade from your vault, open a Bitcoin IRA, set up inheritance solutions. Use code TFTC10 for 10% off a new multisig vault. unchained.com/tftc
• Aven: a Bitcoin-backed line of credit up to $1 million, custodied by BitGo, no rehypothecation, rates from 7.99% APR, fixed for up to 10 years, 2% unlimited cashback. Keep your stack, access liquidity. aven.com/bitcoin
• CrowdHealth: crowdfunded healthcare my family has used for 5 years. Two babies, multiple health events, never going back to insurance. As a family of 5 we pay around $700 a month. Code TFTC gets you $99/month for the first 3 months. joincrowdhealth.com/tftc