h hoge.gg
Subscribe
BTC$67,432.18+2.34%ETH$3,521.44+1.08%SOL$178.62-0.62%BNB$612.30+0.41%XRP$0.6234-0.18%ADA$0.4521+3.12%DOGE$0.1623+1.86%AVAX$38.71-1.24%LINK$17.84+0.92%HOGE$0.00004120+4.21%
BTC$67,432.18+2.34%ETH$3,521.44+1.08%SOL$178.62-0.62%BNB$612.30+0.41%XRP$0.6234-0.18%ADA$0.4521+3.12%DOGE$0.1623+1.86%AVAX$38.71-1.24%LINK$17.84+0.92%HOGE$0.00004120+4.21%
● Mining & Staking

Marathon vs Riot: When Rates, Not Bitcoin, Set the Price

The Fed is set to raise rates for the first time since 2023, and it hits Marathon and Riot harder than Bitcoin does. Both are now capital-intensive builders whose fate turns on the cost of capital.

The decision that outranks Bitcoin’s next move

At 2:00 p.m. ET today, the Federal Reserve is expected to do something it has not done since 2023: raise interest rates. The CME FedWatch tool puts the odds of a quarter-point increase at better than 90 percent going into the meeting, a move that would lift the federal funds target range from 3.50 to 3.75 percent up to 3.75 to 4.00 percent (Coinpedia). For most of the crypto market, the question is simple: what does a hike do to Bitcoin? For the two largest US-listed Bitcoin miners, Marathon (now MARA Holdings) and Riot Platforms, the decision matters for a larger and less obvious reason.

Both companies have stopped being pure bets on the Bitcoin price. Over the past year they have turned into capital-intensive infrastructure builders, each committing to spend billions on power and data centers. When a business has to raise that much money to grow, the price of money becomes the story. The variable that decides their next leg is no longer the hashprice. It is the cost of capital, and today the Fed is about to make it more expensive.

The clearest proof arrived on 18 August, when both stocks fell even as Bitcoin rose, dragged down by a 10-year Treasury yield pushing toward 4.7 percent (24/7 Wall St.). A miner whose shares ignore a Bitcoin rally to follow the bond market is not trading like a miner anymore. It is trading like an infrastructure developer, and that is exactly what these two are becoming.

From Bitcoin proxies to capital-intensive builders

For a decade, owning a mining stock was a leveraged way to own Bitcoin. Revenue rose and fell with the coin, costs were mostly electricity and machines, and the market valued the shares as a high-beta Bitcoin wrapper. That model is breaking on both sides of this comparison, though in different ways.

Riot has signed a 20-year lease to host artificial-intelligence computing for a major AI lab, a contract worth about $9.1 billion at its base term (CoinDesk). Marathon has raised its power target to roughly 4.8 gigawatts and is buying the plants to back it, recasting itself as what it now calls an “energy and digital infrastructure company” (Yahoo Finance). Neither of those is a bet you can price off a single day’s Bitcoin candle.

The catch is that both strategies are enormously expensive up front. A dozen public miners spent $6.87 billion on capital assets in the first half of 2026 alone, more than the $6.50 billion the group spent in all of 2025 (The Energy Mag). Building AI-ready capacity now costs roughly $8 million to $11 million per megawatt once you add liquid cooling and scarce transformers. When your growth plan requires spending at that scale, the interest rate on the money you borrow, and the discount rate the market applies to the cash you promise to earn later, quietly become the most important inputs on the page.

This is not a temporary detour. Both management teams have told investors that the mining business alone cannot carry the company through the next halving in 2028, when the block subsidy falls again and the margin on each coin shrinks. The way out they have chosen, selling compute and power to the AI industry, requires building first and earning later, which is the definition of a capital-intensive business. Once you accept that framing, the Federal Reserve stops being background noise and becomes a direct input into what these shares are worth.

What the Fed is deciding this afternoon

The mechanics are straightforward. The committee has held the funds range at 3.50 to 3.75 percent since December 2025. A quarter-point increase today would move it to 3.75 to 4.00 percent, the first hike of Chair Kevin Warsh’s tenure and the first since 2023. Goldman Sachs dropped its call for a hold in favor of a 25 basis-point rise after August inflation data came in hot, with headline CPI at 3.4 percent and core at 2.4 percent (FinanceFeeds).

The why behind the hike matters for miners too. Inflation has proven sticky rather than spiking, with elevated energy costs keeping the headline number well above target, and the labor market has stayed firm enough to give the committee room to move. That combination, firm growth and stubborn prices, is the classic setup for higher-for-longer, and higher-for-longer is precisely the regime that punishes companies whose value depends on cash promised years from now. For a Bitcoin miner that used to live and die by this month’s hashprice, worrying about the 2028 rate path is a new kind of exposure.

The hike itself is close to fully priced, so the reaction will hinge on the parts that are not: the Summary of Economic Projections and the dot plot that accompany this meeting. In June, the median official already saw the funds rate reaching 3.75 to 4.00 percent by December, so a move today merely completes that path; the real signal is whether the dots now point higher for longer, and whether Warsh treats October as a live meeting (Yahoo Finance). Benzinga framed the dot plot and the growth and unemployment forecasts as the three things worth watching precisely because the hike is a foregone conclusion (Benzinga).

Crypto goes into the decision already bruised. The CLARITY Act, the market-structure bill the industry spent two years pushing, failed its Senate cloture vote on 15 September, falling short of the 60 votes needed and effectively dying for 2026 (The Crypto Times). With the regulatory catalyst gone and a hike incoming, Bitcoin slid toward the mid-$70,000s. We mapped how these two clocks, the legislative one and the monetary one, were always going to collide in our look at crypto’s September countdown.

Why a rate hike lands hardest on a miner turned builder

There are three channels through which higher rates reach a company like Riot or Marathon, and they explain why the market now watches the bond desk instead of the mempool.

The first is the discount rate. A data-center lease that pays rent every year until 2048 is, in valuation terms, a very long bond. The further into the future a cash flow sits, the more its present value shrinks when the rate you discount it at goes up. Long-duration promises are the most rate-sensitive assets there are, which is why a contracted developer can lose value on a hawkish dot plot even if not a single tenant misses a payment.

The math is not subtle. Picture a fixed payment arriving two decades from now. Discounted at 8 percent, a dollar of that future rent is worth about 21 cents today; lift the discount rate to 9 percent and it is worth closer to 18 cents, a drop of roughly a sixth from a single percentage point. Stretch that across twenty years of contracted payments and a modest move in rates can swing a lease-based valuation by double-digit percentages, which is why a company selling 20-year cash flows behaves, on the stock screen, like a long-dated bond rather than a Bitcoin miner.

The second is the cost of new money. Both companies still have to fund construction that is years from completion, and every dollar of that build is financed with debt or equity; a higher policy rate raises the coupon on the bonds and the hurdle on the shares. The third is the opportunity cost of Bitcoin itself. A coin that pays no yield competes with cash, and when cash pays 4 percent the marginal buyer demands more from a non-income asset, which is part of why higher yields pressure the Bitcoin price and, with it, any balance sheet stuffed with coins. The self-correcting floor under mining, where higher-cost rigs power down and difficulty falls, still exists, a dynamic we traced in our review of the hashrate stall a year after the zettahash, but it does nothing to shield a company from its own financing costs.

The Q2 2026 scorecard

Before the strategy, the numbers. Both companies reported second-quarter results in early August, and the two income statements already look like different businesses.

Metric (Q2 2026)MARA Holdings (MARA)Riot Platforms (RIOT)
Total revenue$174.9M (-27% YoY)$174.2M (+14% YoY)
Net loss-$611.3M (-$1.60/sh)-$237.2M (-$0.68/sh)
Adjusted EBITDA-$360.9M-$69.7M
Non-mining revenueNot material~$60.5M (data center + engineering)
Bitcoin produced2,422 BTC1,587 BTC
Hashrate70.3 EH/s energized44.4 EH/s deployed
Bitcoin held (30 Jun)35,577 BTC11,380 BTC (5,821 pledged)
Anchor AI tenantNone at scaleAnthropic, 191 MW, 20 years
CEOFred ThielJason Les

The headline is the shape of the loss, not just its size. Marathon’s $611 million loss was driven mostly by a $343 million unrealized markdown on its Bitcoin, an accounting swing that reverses when the coin rallies (SEC 8-K). Riot’s smaller loss came alongside its first material stream of non-mining revenue, roughly $60 million from its data-center and engineering segments (Riot Platforms). One company is still a Bitcoin balance sheet that mines; the other is starting to look like a landlord that also mines.

Riot: a 22-year rent check that trades like a long bond

Riot’s transformation is the cleaner story, and it is the one that makes the rate decision so pointed. Its lease with the AI lab, reported to be Anthropic though Riot officially calls it a “leading frontier AI lab,” runs 20 years through June 2048 for 191 megawatts of critical IT capacity at its Rockdale, Texas campus (DataCenterDynamics). The base term is worth about $9.1 billion, rising to as much as roughly $16 billion if two five-year extensions are exercised, at gross margins the company pegs in the mid-80s percent. Riot plans to deliver 96 megawatts by December 2027 and the full 191 by June 2028.

That is a bond-like cash-flow stream, and the market is learning to value it like one. Chief executive Jason Les called the deal “a defining moment in our evolution into a leading developer of large-scale data centers” (Riot Platforms). The activist investor Starboard Value had pushed exactly this pivot in February, arguing Riot’s power capacity could be worth billions more redirected to AI (CoinDesk). The trade-off is duration risk. A stream of contracted rent to 2048 is precisely the kind of asset a higher-for-longer dot plot punishes, because its entire value sits in distant years. Riot has also gone deliberately Bitcoin-light, holding 11,380 coins with 5,821 of them pledged as collateral, so it carries far less of the treasury exposure that whipsaws Marathon.

The lease is also not Riot’s only contracted line. A separate deal with AMD adds 50 megawatts, taking total contracted capacity to about 241 megawatts and lifting combined contracted revenue toward roughly $9.8 billion, while the Anthropic agreement alone is projected to throw off net operating income of about $365 million to $411 million a year once fully delivered (Riot Platforms). To bridge construction before that rent starts flowing, Riot lined up a $573 million interim facility from Morgan Stanley, the kind of financing that gets more expensive to roll over the higher the Fed pushes rates. That is the tension in a single sentence: the contract is a long-dated asset, but the money to build it is short-dated and floating.

Marathon: a leveraged Bitcoin balance sheet meets a rate hike

Marathon is the mirror image. It still holds 35,577 Bitcoin, a treasury worth well over $2 billion that it marks to market every quarter, which is why its earnings swing violently with the coin (SEC 8-K). Rather than sign a single anchor tenant, chief executive Fred Thiel has pursued vertical integration by owning the power. The company raised its portfolio target to about 4.8 gigawatts and agreed to buy the Long Ridge combined-cycle gas plant in Ohio for roughly $1.5 billion, alongside a joint venture with Starwood aimed at gigawatt-scale capacity.

Thiel’s thesis is that energy, not chips, is the moat. “By 2028, you’ll either be a power generator, be owned by one, or be partnered with one,” he told CoinGeek. “The days of being a miner plugged into the grid are numbered” (CoinGeek). He also describes mining bluntly as “a zero-sum game,” one where “the floor is your energy cost.” The problem for Marathon on a day like today is that its model exposes it to rising rates on two fronts at once. Its Bitcoin treasury is pressured when higher yields sap demand for a no-yield asset, and its buildout still lacks the contracted, investment-grade cash flow that would let it finance cheaply. It has been funding the pivot partly by selling coins, part of a sector-wide move to raise cash for AI (CoinDesk), and partly with convertible notes whose refinancing math gets harder as rates climb.

Marathon has leaned into the identity shift, describing itself now as a company that converts excess energy into digital capital rather than a pure miner (Yahoo Finance). The operating numbers show why the label matters. It reported a purchased-energy cost of $38,690 per Bitcoin in the second quarter, a figure that looks far better than Riot’s, but it is a narrow measure built on cheap owned power and is not comparable to Riot’s all-in cash cost of $49,912 or its fully loaded cost above $90,000 per coin (Riot Platforms). The point for an investor is that Marathon’s edge on the electricity bill is real, but it has yet to convert that edge into the contracted, bankable cash flow that earns a low cost of capital, and until it does, the rate cycle works against it.

Cost of capital is the new hashprice

Here is the twist that makes today’s meeting matter. The AI pivot did not just give miners a new revenue line; it lowered their cost of capital. As these companies signed long contracts with creditworthy tenants, lenders re-rated them from speculative crypto plays toward infrastructure credits. Bond deals that cleared above 9 percent in 2025 were landing closer to 6 to 7 percent in 2026 across names like Core Scientific, Cipher, Applied Digital and Hut 8 (Miner Weekly). Cheaper money is the whole game when you are spending billions on data centers.

Riot and Marathon are not doing this alone. The public mining sector has announced more than $70 billion in cumulative AI and high-performance-computing contracts, and the scoreboard of who has signed what has become the main way the market sorts winners from laggards (CoinShares). The table below shows where the two protagonists sit among their peers.

CompanyHeadline AI/HPC contractApprox. value
Core ScientificAMD colocation (529 MW) and CoreWeave~$14B and over $10B
TeraWulfHPC hosting contracts~$12.8B
Riot PlatformsAnthropic lease (191 MW) plus AMD (50 MW)$9.1B base (~$16B max)
Hut 815-year data-center lease (352 MW)~$9.8B
MARA HoldingsNo anchor tenant signed yetNot contracted

A rate hike, and especially a dot plot that signals more to come, leans against that tailwind. The sector still faces an enormous funding requirement. Bernstein has estimated a sector-wide funding gap around $50 billion, warned that returns on capital are likely single digits for most of these deals, and cautioned that the eye-popping numbers at Core Scientific’s CoreWeave arrangement, an unusually capital-light structure, are the exception rather than the rule; stabilized returns nearer 4 to 5 percent on assets at TeraWulf and Cipher are a better guide to what a miner-turned-landlord should expect (Bernstein via MSN). When your expected return on a project is a single-digit percentage, the rate you pay to fund it is not a detail. It is the difference between a good deal and a bad one. This is the same logic that governs any business built on committed capital, from data centers to the validators we examined in our study of validator economics and the cost of locked-up ETH.

The buildMARA HoldingsRiot Platforms
Headline project~4.8 GW power pipeline; Long Ridge 505 MW gas plant; Starwood JV191 MW AI fit-out for Anthropic at Rockdale; Corsicana expansion
Estimated capexMulti-billion (Long Ridge alone ~$1.5B)~$2.1B to $2.3B for the Anthropic build (Bernstein est.)
Contracted revenueNone at scale$9.1B base, up to ~$16B with two 5-yr options
Cash-flow durationShort (spot mining + mark-to-market treasury)Long (contracted rent to June 2048)
Main funding toolsConvertible notes, at-the-market equity, Bitcoin sales$573M Morgan Stanley interim facility, project finance, equity
Bitcoin on balance sheet35,577 BTC11,380 BTC (5,821 pledged)
Key overhangNo anchor tenant signed yetSingle-tenant concentration + delivery timeline

How the rate decision transmits to each

The two companies are exposed to the same hike through different doors. The table below lays out the channels and how each one lands.

ChannelEffect on MARAEffect on RIOT
Discount rate on contracted cash flowsSmall (little contracted revenue)Large (long-duration rent value falls as rates rise)
Cost of new debt for capexHigher coupons on future notes and convertsHigher cost to term out interim facility and project debt
Equity dilution riskHigh (funds buildout via share issuance)Moderate (some contracted cash to lean on)
Bitcoin-treasury markLarge (35,577 BTC; BTC pressured by higher yields)Small (Bitcoin-light by design)
Refinancing exposureConvertibles due 2030 and 2031Interim facility to be refinanced into long-term debt
Net effect of a hawkish dot plotDouble hit: Bitcoin price plus funding costDuration hit: rent value plus capex cost

The summary is uncomfortable for both, but for opposite reasons. Marathon takes the more direct blow to its Bitcoin, while Riot takes the subtler blow to the present value of its 22-year rent check. Neither is the classic mining risk of a bad hashprice week.

The Bitcoin leg still matters, just less and differently

None of this means Bitcoin stopped mattering. It means the two companies now carry very different amounts of it. Bitcoin traded around $76,000 to $77,000 into the decision, roughly 40 percent below its October 2025 record near $126,000 (FinanceFeeds). JPMorgan’s Nikolaos Panigirtzoglou has pegged the industry’s all-in production cost near $78,000, above the current price, and describes the self-correcting mechanism this creates: “When bitcoin trades below its production cost, higher-cost miners power down, the hashrate declines, and difficulty adjusts lower” (TFTC).

The accounting makes the exposure sharper. Under current fair-value rules, both companies mark their Bitcoin to market every quarter, so a falling coin drops straight to the bottom line as an unrealized loss and a rising one becomes an unrealized gain. That is why Marathon’s $343 million second-quarter markdown can reverse in a single good quarter, and why its reported earnings tell you as much about the Bitcoin chart as about the mining business itself. Riot, holding a far smaller stack, feels little of that swing, which is part of what lets its shares trade on the lease rather than on the ticker.

For Marathon, with more than 35,000 coins on the books, every move in Bitcoin flows straight through earnings and net asset value. For Riot, holding under a third as many, the coin is a smaller factor than the discount rate applied to its lease. There is a demand signal underneath the price weakness, though: US spot Bitcoin ETFs took in $159.9 million on 14 September, led by BlackRock’s IBIT at $134.3 million (Coinpedia). The steady institutional bid those products represent is part of the same story we told in our guide to crypto ETF approvals and the commodity-or-security gate.

What Wall Street will pay for each

The divergence shows up most clearly in analyst price targets, which have split into two camps. Bernstein’s Gautam Chhugani rates Riot Outperform with a $35 target, having rebuilt the model to include the AI infrastructure business, while rating Marathon only Market Perform at $17 (The Block). Citi has Riot at Buy with a $32 target. Marathon draws a wider and lower spread: a consensus in the high teens, but with Morgan Stanley at $6 Underweight and JPMorgan reportedly down at $13 (StockAnalysis).

Read those numbers against the tape and the message is plain. Riot trades near $21 for a market value close to $8 billion; Marathon trades near $11 for about $4.3 billion, inside a 52-week range of $6.66 to $23.45. The market is paying up for contracted, infrastructure-style cash flow and marking down an unproven, Bitcoin-heavy option, even though Marathon owns far more coins and more raw power ambition. That is the rate regime talking. As one markets analysis put it after the August sell-off, contracted AI capacity, secured power and the cost of capital now set these valuations, not the size of the Bitcoin treasury (24/7 Wall St.).

It is worth being clear about what this does and does not say. A rate hike is not good for either stock; both are capital-hungry, and tighter money raises the bar for each. The claim the market is making is narrower and about relative positioning: given a rising cost of capital, investors would rather own the company whose cash flows are already contracted and underwriteable than the one still promising to build and still hoping to sign. That preference can reverse quickly if the dot plot turns dovish or if Marathon lands the anchor tenant it has been chasing, which is why both names carry outsized single-day moves around meetings like this one.

The risks a cheap rate used to hide

A low cost of capital papers over a lot of risk. As rates rise, three of those risks come back into focus. The first is concentration. Riot’s contracted revenue rests heavily on a single frontier AI lab; if that tenant’s plans change, the bond-like stream that justifies the valuation looks a lot less certain. The second is obsolescence. The buildings can last decades, but the GPUs inside them age in a few years, and who eats that hardware risk depends entirely on how the contract is written. The third is execution. Riot must deliver 96 megawatts by the end of 2027 and 191 by mid-2028, and Marathon must actually sign the anchor tenant it still lacks while financing 4.8 gigawatts of power.

Then there is the refinancing wall. Marathon has convertible notes coming due in 2030 and 2031 that will have to be rolled over or repaid, and if the coin and the share price are soft when that day arrives, refinancing in a higher-rate market is exactly when a leveraged balance sheet hurts most. Riot faces a gentler version of the same problem: its $573 million interim facility is meant to be replaced by cheaper long-term debt once the Anthropic build is generating rent, and every notch higher in rates raises the cost of that permanent financing. A sector-wide funding gap estimated near $50 billion means both companies are competing for capital in the same crowded market, at the very moment the Fed is making that capital scarcer.

Bernstein’s caution about single-digit returns matters here, because a project that pencils at a 5 percent return can be sunk by a cost of capital that drifts up to meet it. James Butterfill of CoinShares has argued that some listed miners could draw as much as 70 percent of revenue from AI by the end of 2026, up from roughly 30 percent, while calling this one of the most challenging periods the industry has faced (The Block). The pivot is real, but it is being financed into a rising-rate wind.

The SEC backdrop and what a failed CLARITY vote leaves behind

Both companies file with the US Securities and Exchange Commission as ordinary public issuers, and their mining itself sits on relatively settled ground: the SEC’s Division of Corporation Finance said in March 2025 that proof-of-work mining on public networks is not the offer or sale of a security (SEC). What is not settled is the market-structure question CLARITY was meant to answer, namely which digital assets are commodities and which are securities, and which agency polices them. With the bill dead for 2026, that ambiguity persists, and it is one more reason capital-markets access, not token classification, is where these miners now compete.

There is a quieter irony in the pivot, too. By leasing capacity to an AI tenant, Riot converts an uncertain, Bitcoin-linked revenue line into a contracted one a bank can underwrite. That is a financial de-risking as much as a commercial one, and it is exactly why the market rewards it with a lower cost of capital, at least until the Fed raises the price of money for everyone.

What to watch after 2 p.m. ET

The hike is the easy part. The harder read comes in the projections and the press conference. If the dot plot pushes the expected path higher and Warsh signals that October is a live meeting, long-duration assets, very much including Riot’s lease-based valuation, take the brunt. If the committee frames today’s move as the last step in a path it already sketched in June, the relief could be sharp, especially for the rate-sensitive names.

For the companies themselves, the milestones are concrete. Watch Riot’s delivery cadence toward 96 megawatts by December 2027, and watch whether Marathon can finally announce an anchor tenant at scale to convert its 4.8-gigawatt ambition into contracted cash flow. Both are still, at heart, in the business of turning electricity into something valuable. What changed in 2026 is that the market decided to price them on how cheaply they can fund that conversion. When companies borrow against their coins to build, as both effectively do, the interest rate is everything, a point we made in plainer terms in our practical guide to borrowing against crypto. Today the Fed sets that rate, and for Marathon and Riot alike, that is the number that counts.

Frequently Asked Questions

Is the Fed raising interest rates on 16 September 2026?

The Federal Reserve is expected to raise the federal funds target range by a quarter point at its 16 September 2026 meeting, to 3.75 to 4.00 percent, in what would be its first increase since 2023. Markets price the move at better than 90 percent ahead of the 2:00 p.m. ET decision, so the bigger question is the forward path shown in the dot plot rather than the hike itself.

Why do Bitcoin miners care so much about interest rates now?

Marathon and Riot have become capital-intensive builders that must raise billions to expand into power generation and AI data centers. Higher rates increase the cost of the debt and equity that fund those projects, and they lower the present value of long-dated contracted revenue, so the cost of capital now drives their share prices more than the Bitcoin price does.

What is the difference between Marathon and Riot in 2026?

Riot has signed a 20-year lease to host AI computing worth about $9.1 billion and holds relatively little Bitcoin, so it trades like a contracted infrastructure developer. Marathon holds more than 35,000 Bitcoin and is building roughly 4.8 gigawatts of owned power without a signed anchor AI tenant, so it trades like a leveraged Bitcoin balance sheet with an unproven infrastructure option.

Who is Riot’s AI data-center tenant?

Riot officially describes the counterparty as a leading frontier AI lab and has not named it, but multiple outlets including CNBC and DataCenterDynamics have reported it is Anthropic. The lease covers 191 megawatts at Riot’s Rockdale, Texas campus and runs for 20 years, with two five-year extension options that could lift its total value toward $16 billion.

Is Marathon or Riot the better buy after the Fed decision?

Analysts are split. Bernstein rates Riot Outperform with a $35 target and Marathon only Market Perform at $17, reflecting a preference for contracted cash flow over Bitcoin-heavy optionality, while Marathon’s targets range from $6 at Morgan Stanley to the high teens. This article is analysis, not investment advice, and both stocks are highly sensitive to rates and the Bitcoin price.

By Yuki Tanaka, senior markets writer at HOGE Wire.

Share 𝕏 Post Telegram