JPMorgan, through Nikolaos Panigirtzoglou's team, has published one of the most explicit theses of the year on Ethereum: ETH will continue to lag behind BTC unless on-chain activity, DeFi, and RWA rebound. They back this with hard data: L1 gas fees plummeted 95% year-over-year to an average of 0.5 gwei, rollup revenue to the base layer fell 90% (from $113 million to $10 million between 2024 and 2025), ETH's circulating supply growing at a net 0.23% annually, and an ETH/BTC ratio of 0.02835, a ten-month low. The spot price of ETH at $2,113 and the total crypto market capitalization at $2.56 trillion serve as the backdrop, not the thesis. This article dissects exactly what JPMorgan is saying, what supports it, what contradicts it, and what a traditional bank misses when looking at Ethereum through a DCF lens.
Panigirtzoglou's report is not a short-term bearish price target for ETH. It is a thesis on structural relative underperformance against BTC. The distinction is important because it forces a discussion of economic mechanics (what captures value in a base layer) and not just flows. In this analysis, we will go into detail: the three explicit conditions JPMorgan sets as triggers to reconsider its thesis, how the EIP-1559 equilibrium was broken with L2 maturity, which on-chain metrics support and which challenge the reading, and why banking analytical frameworks may be looking at Ethereum as a software equity when it more closely resembles a monetary asset with demand for economic security.
Quick Glossary to follow the article:
- L2 Rollup (Layer 2): A secondary network built on top of Ethereum (mainnet or L1) that bundles thousands of transactions off-mainnet, executes them more cheaply, and only publishes a verifiable summary on L1. Examples: Arbitrum, Optimism, Base, zkSync, Scroll. They reduce transaction costs by 10-100x compared to using Ethereum directly.
- Mainnet / L1: The base Ethereum network, where all security resides and where L2 rollups write their summarized data.
- EIP-1559: The rule, in effect since 2021, that burns (destroys) part of the fees from each transaction on L1. If L1 activity decreases, burning decreases, and ETH ceases to be deflationary.
- Blob: The new, cheaper format (introduced by EIP-4844 in 2024) that L2 rollups use to upload their data to L1. They pay cents instead of dollars, but L1 earns much less.
Editorial Disclaimer: This article is for informational purposes only and does not constitute financial advice or recommendation. Prices and flows change daily. Data as of May 2026.
What exactly is JPMorgan's thesis on Ethereum?
Nikolaos Panigirtzoglou, Managing Director of JPMorgan's Global Markets Strategy team, authored a research note for institutional clients, asserting a specific position: Ethereum and the broader altcoin market will continue to underperform Bitcoin unless three simultaneous changes materialize on the network. This is not a sell rating, nor is it a price target. It is a thesis on the persistence of relative underperformance.
The very title of the note — cited by Bloomberg — frames the debate as a question of value capture, not technology. JPMorgan explicitly acknowledges that Ethereum is the most demanded base layer of Web3, hosting most of the DeFi infrastructure, and is the chosen standard for tokenized fund issuers like BlackRock's BUIDL or iShares' ETHB with staking. The problem, according to the bank, is that this technical centrality is not translating into persistent demand for the ETH token.
JPMorgan deploys four quantitative supports:
- The ETH/BTC ratio falls to 0.02835, a ten-month low and a 35% retreat from the August 2025 peak of 0.04324.
- US spot BTC ETFs have recovered two-thirds of previous outflows following the correction due to geopolitical crisis; ETH ETFs only one-third.
- Open interest in CME BTC futures has almost returned to previous highs; ETH futures remain clearly depressed.
- Median L1 gas fees fall 95% year-over-year to 0.5 gwei, neutralizing the EIP-1559 burning mechanism.
Today's market reading reinforces this framing: ETH opens at $2,113, a six-week low, while global crypto market capitalization loses $180 billion in the last week to $2.56 trillion. To place this point in the context of the other angle, we recommend our analysis BTC vs ETH: ETF Flow Divergence in May 2026, which examines daily flows and exchange-traded products in detail. Here, however, we focus on the economic mechanics of the banking thesis.
What are the three conditions JPMorgan requires to revise its thesis?
Panigirtzoglou is unusually explicit about the triggers that would invalidate the thesis. He doesn't speak in abstracts: he sets three conditions that must be met simultaneously for the note to shift towards a neutral or constructive outlook on ETH.
| JPMorgan Condition | Proposed Objective Metric | Current Value | Suggested Reversal Threshold |
|---|---|---|---|
| L1 On-chain Activity Rebound | Average L1 Gas (gwei) | 0.5 | 5-10 sustained |
| DeFi Revitalization on Ethereum | Mainnet TVL (B$) | 45.4 | 65-75 |
| RWA Adoption and Real Applications | Tokenized RWA on ETH (B$) | ~9.2 | 25-30 |
The first condition is the most mechanical. For the EIP-1559 burn to once again act as a brake on issuance, and thus for the scarcity narrative to apply, the L1 base fee must consistently remain above levels where daily burning exceeds consensus issuance. JPMorgan does not set an exact number, but its economists suggest a range between 5 and 10 gwei as the necessary floor, compared to the current 0.5 gwei.
The second condition requires a reactivation of DeFi on mainnet, not on L2. This is an important nuance: the bank acknowledges that the aggregated TVL of the Ethereum + L2s ecosystem has grown, but argues that it is on mainnet where gas is burned and natural buying pressure on ETH is generated. A return to levels near $65-75 billion in L1 TVL — last seen during the peak of the 2021 cycle — would be the clear signal.
The third condition, regarding RWA, is the most interesting because it is the one JPMorgan is closest to seeing fulfilled. The tokenization of money market funds, sovereign debt, and private credit has moved from concept to operational reality with BUIDL, BENJI, ONDO, and several European bank pilots. But the bank insists that current volumes — about $9.2 billion of RWA tokenized on Ethereum according to rwa.xyz — are still too small to move the needle on fee flow and token value capture.
Why are L2 rollups technically a success but economically a problem?
The technical core of JPMorgan's thesis lies here. When Ethereum implemented EIP-4844 (Proto-Danksharding) in 2024, the goal was to drastically reduce the cost for rollups to publish data availability on mainnet. It succeeded. Rollup transaction costs fell by 10 to 100 times. Transactions on Arbitrum, Base, or Optimism began to cost fractions of a cent. This is, without a doubt, one of the greatest scalability achievements in the protocol's history.
The collateral effect is severe: if rollups pay less per block, mainnet collects less. And mainnet today receives the bulk of its demand precisely because it is the settlement layer for rollups, not for end-users. The consequence is seen in a single figure: the aggregated payment from rollups to the L1 network fell from $113 million in 2024 (41% of L2 operators' total revenue) to $10 million in 2025. A 90% year-over-year drop.
| Period | Average L1 Gas (gwei) | Average L2 Swap Cost ($) | Fees Paid by L2 to L1 ($M) | Δ ETH Supply (annualized) |
|---|---|---|---|---|
| 2021 (DeFi/NFT peak) | 80-150 | N/A | N/A | +4.3% |
| 2023 (post-Merge) | 20-40 | 0.30-1.00 | N/A | -0.2% |
| 2024 (post-Dencun) | 5-15 | 0.02-0.10 | 113 | -0.05% |
| 2025 | 1-3 | 0.01-0.05 | 10 | +0.15% |
| May 2026 | 0.5 | 0.001-0.01 | ~3 (run-rate) | +0.23% |
The mechanical translation is clear. EIP-1559 stipulates that the base fee of each transaction is burned (removed from circulating supply) instead of going to validators. The net supply balance is: consensus issuance minus base burn. When burning falls to near-zero levels, staking issuance dominates the balance, and supply grows net. This is exactly what is happening: ETH is expanding at a net 0.23% annually instead of contracting as it briefly did in 2023.
If you want to see the mechanics in detail, we recommend our analysis of Geoff Kendrick's (Standard Chartered) thesis on ETH at $7,500, which argues the opposite of JPMorgan: that value capture will migrate to staking and re-staking, not fees, and that this is enough to sustain structural appreciation.
How was the ultrasound money narrative broken?
During 2022-2023, after the merge to proof-of-stake and under conditions of high on-chain activity, ETH experienced several quarters of negative net supply change. EIP-1559 burning was greater than staking issuance. The term "ultrasound money" — scarcer than Bitcoin's "sound money," which only disinflates through halvings — became popular in the community. It was a powerful monetary argument because it tied network utility (more use = more burning) to asset scarcity.
That link has been broken. Not by a protocol decision, but by an architectural decision: to push volume to L2 and reduce data availability costs. The net daily burn of ETH in May 2026 is around 1,200-1,800 ETH, compared to the 8,000-12,000 ETH daily seen during periods of high activity in 2023. The exact number can be checked in real-time on ultrasound.money, a dashboard maintained by the community itself.
Is it solvable without abandoning L2s?
There are three paths discussed in EthResearch and core forums. The first is to raise the L1 gas limit to absorb activity that currently goes to L2 (Glamsterdam line with EIP-7732). The second is to reorganize blob economics so that the price floor is higher even with low demand (proposals under study for Hegota). The third is to assume that value capture permanently shifts to staking and re-staking (EigenLayer, Symbiotic) and stop measuring ETH by DCF of fees. Each path has its defenders; JPMorgan remains skeptical of all three because none are implemented and because fresh demand is not appearing.
What do DeFi metrics say about the health of the ecosystem?
Ethereum mainnet's market share in global DeFi TVL has dropped from 63.5% at the beginning of 2025 to 54% in May 2026, with an absolute value of $45.4 billion on L1 according to DefiLlama. The decline is not explained by an outflow, but by the specialization of alternative networks that have captured specific niches.
| Network | DeFi TVL Share (%) | TVL (B$) | Dominant Niche |
|---|---|---|---|
| Ethereum L1 | 54.00 | 45.40 | Institutional settlement, RWA, money markets |
| Solana | 6.66 | 5.60 | Fast execution, memecoins, perp DEX |
| BNB Chain | 6.60 | 5.55 | Retail onboarding via Binance |
| Bitcoin (BTCFi) | 6.35 | 5.34 | BTC collateral on native L2s |
| Tron | 6.17 | 5.19 | Stablecoins (USDT 97.86% of stock) |
| Base (L2 OP Stack) | 5.44 | 4.58 | Coinbase retail onboarding |
| Hyperliquid | 1.81 | 1.52 | Perp DEX, synthetic derivatives |
What JPMorgan emphasizes is not the loss of market share — a predictable and healthy phenomenon of a specializing ecosystem — but that this capital outflow is not returning in the form of new L1 demand. Solana has consolidated the retail speed segment. Tron is the de facto network for global USDT with $89.6 billion in stablecoins. BTCFi emerged in 2026 as the surprise of the year by capturing 6.35% of TVL by leveraging Bitcoin collateral. Each of these niches was theoretically capturable by Ethereum mainnet with the right architecture; none has been.
Additionally, the bank points to three factors limiting institutional investment in altcoins in general:
- Lower market depth and higher slippage during liquidity events.
- Stagnation of retail DeFi: the number of unique monthly active addresses on mainnet has been flat for three quarters according to Etherscan.
- Continued security incidents: bridge hacks, lending exploits, and, in 2026, the case of the Aave 200k governance vote which showed tensions in cross-chain management.
Do Glamsterdam and Hegota solve the problem JPMorgan diagnoses?
Ethereum's roadmap has concrete technical answers to some of the problems JPMorgan points out, but the bank dismisses them as insufficient as long as they are not implemented and as long as it is not proven that they generate new demand, not just new capacity.
| Upgrade | Window | Core EIPs | L1 Gas Limit (M) | Economic Focus |
|---|---|---|---|---|
| Glamsterdam | Q3 2026 | EIP-7732 (ePBS), EIP-7928 (BALs), EIP-8037 | 200 | 5x capacity, parallelization, in-protocol MEV |
| Hegota | Q4 2026 / H2 2026 | Verkle Trees, EIP-7805 (FOCIL), EIP-8141 (smart accounts) | 200 | Stateless nodes, account abstraction, censorship resistance |
Glamsterdam attempts to solve three distinct problems in a single update. First, ePBS (EIP-7732) brings the block auction market into the protocol, eliminating external relays like Flashbots and making the flow more auditable — a clear nod to institutional compliance. Second, Block-Access Lists (EIP-7928) allow parallel processing of transactions without state conflicts, which theoretically multiplies effective block throughput. Third, EIP-8037 adjusts state storage costs to prevent a 200 million gas limit from exploding the node database.
Hegota is technically more radical. Verkle Trees replace the Merkle Patricia structure and reduce the size of proofs a validator must process by 90%, opening the door to stateless clients: nodes that don't need to download gigabytes of history to sync. EIP-7805 (FOCIL) strengthens censorship resistance by assigning random participants the ability to enforce pending transactions in blocks. And EIP-8141 brings native smart accounts with multisig and fee sponsorship.
Here's the economic question JPMorgan poses without answering: what happens if Ethereum quintuples its L1 capacity and no fresh demand appears to fill those blocks? The mechanical answer is that the base fee tends even more towards zero. Glamsterdam could, paradoxically, worsen the value capture problem in the short term if it is not accompanied by new use cases anchored in mainnet. The bank does not rule out their appearance — it just says it doesn't see them today.
What traditional banks miss in their analysis?
This is where JPMorgan's thesis becomes debatable. Banking analytical frameworks for valuing blockchain assets are inherited from software company valuation: discounted future cash flows, revenue multiples, interest rate sensitivities. Applied to ETH, the natural result is what Panigirtzoglou says: if fees fall, the asset depreciates. But ETH is not just a software company with tokens in circulation. There are at least three dimensions that the banking framework poorly captures.
Staking as primary yield, not fees
Approximately 37 million ETH (30% of the circulating supply of 120 million) are locked in staking contracts. These validators receive consensus issuance, not just fees. The real yield perceived by an ETH staker comes from three blocks: consensus issuance, transaction tips, and MEV. EIP-1559 burning is an adjustment to supply, not a component of validator yield. If a tokenized fund like iShares' ETHB commits between 70% and 95% of its assets under management to staking, it does so for the yield, not for deflation. This is an input that JPMorgan poorly models because its benchmark is a bond coupon, not the yield of a base layer.
Re-staking and the economic security market
EigenLayer and Actively Validated Services (AVS) have created a market where staked ETH can be lent as economic security collateral for external services: data availability, oracles, bridges, sequencers. The aggregated yield of staked plus restaked ETH surpasses pure staking and creates structural demand for ETH as a security asset. This is a component that traditional DCF does not capture because it does not appear on the "Ethereum mainnet" income statement.
RWA and stablecoins: the de facto settlement base
The stock of stablecoins issued on Ethereum mainnet (USDC, USDT, PYUSD, various GENIUS-compliant tokens) exceeds $95 billion, and BlackRock's BUIDL has accumulated over $2.9 billion in assets. Every transfer, redemption, or issuance of a tokenized financial instrument pays fees in ETH. It is a silent but growing flow, linked to post-CLARITY Act regulation that we covered in our analysis of the CLARITY Act and the ban on stablecoin yields. If RWA flows increase from $9.2 billion to $30-50 billion in 18 months — something not unreasonable if large asset managers continue their tokenization program — the pressure on L1 fees would materially change.
What concrete signals would trigger a change in thesis?
If the Panigirtzoglou thesis is disproven, it will be through four observable metrics. Each reader can monitor them independently without relying on the bank's or our editorial line.
| Indicator | Source | May 2026 Reading | Thesis Change Threshold |
|---|---|---|---|
| Average L1 Gas (gwei) — 30d average | Etherscan | 0.5 | ≥5.0 |
| Net Daily ETH Burn | ultrasound.money | 1,500 | ≥5,000 |
| Mainnet DeFi TVL (B$) | DefiLlama | 45.4 | ≥65 |
| Tokenized RWA on Ethereum (B$) | rwa.xyz | 9.2 | ≥25 |
| Net 30d Flows to Spot ETH ETFs (M$) | SoSoValue, Farside | ~-220 (cumulative) | +500 sustained |
| ETH/BTC Ratio | CoinGecko | 0.02835 | ≥0.038 |
The first three metrics are JPMorgan's own explicit agenda. The last three are the aggregated markers the market watches to validate the narrative. The combination matters: two metrics can move upwards due to temporary catalysts (a retail rally, a tactical rotation) without the thesis structurally changing. At least four of the six are needed for a bank like JPMorgan to be forced to rewrite the note.
How does this reading fit with other institutional catalysts in 2026?
The JPMorgan thesis is not published in a vacuum. There are other ongoing institutional catalysts that should be contextualized to avoid binary readings.
On the bearish side, in addition to the Panigirtzoglou thesis, the macro environment of long-term high interest rates pressures all high-duration assets — and ETH, modeled as a software asset by the banking framework, is one. Institutional risk aversion after the correction due to the Hormuz conflict favors Bitcoin as a reserve, a phenomenon we covered in detail in Bitcoin's institutional accumulation versus the retail sell-off in April.
On the bullish side, there are three counteracting flows. First, ETFs with staking (Grayscale Ethereum Staking ETF launched on April 6 on NYSE Arca, iShares' ETHB) capture ETH for staking in proportions of 70-95% of assets under management, removing the asset from free float. Second, BUIDL and tokenized money market funds continue to grow: BlackRock's BUIDL surpassed $2.9 billion, and European issuers are preparing similar launches. Third, Geoff Kendrick's (Standard Chartered) optimistic note on ETH at $7,500 in 2026 offers the opposite reading to JPMorgan, using the same on-chain data but different weights in the model.
Institutional consensus does not exist. What exists is a market where the ETH/BTC ratio falls to 0.02835, capital currently prefers the asset with greater narrative simplicity (BTC), and honest analysts can defend opposing theses with the same data. For the structural investor, the useful reading is: the JPMorgan thesis is not the truth, it is a falsifiable hypothesis whose falsification conditions are observable. Monitoring the six metrics in the table above is more useful than taking sides a priori.
Key takeaway for the reader: JPMorgan's thesis, signed by Panigirtzoglou, is a falsifiable statement, not a verdict. It states that ETH will lag behind BTC unless on-chain activity, DeFi, and RWA rebound. If your thesis on ETH rests on any of these three factors, the practical question is not whether JPMorgan is right, but what concrete thresholds for L1 fees (≥5 gwei sustained), mainnet TVL (≥65 B$), RWA (≥25 B$), and ETF flows (+500 M$/30d) you would consider sufficient to confirm or reject your view. The weakness of traditional banking analysis is that it applies a software DCF framework to an asset that combines properties of a base layer, money, and an economic security market. The weakness of crypto-native analysis is that it tends to dismiss value capture because "the protocol is what matters." The useful insight lies in the middle: measuring what both parties agree is measurable, without narrative shortcuts.
Are you positioned in ETH or BTC? Do you know what you hold in DeFi on each network? CleanSky — your banking app for DeFi — shows your DeFi portfolio across more than 484 protocols and over 34 networks, including Ethereum mainnet, Arbitrum, Base, Optimism, and new L2s using blobs. Paste any address: check positions, risk scores, and token approvals. No sign-ups, no wallet connection.
Frequently Asked Questions
What exactly does JPMorgan's report say about Ethereum?
The team led by Nikolaos Panigirtzoglou at JPMorgan argues that Ethereum and altcoins will continue to structurally lag behind Bitcoin unless three simultaneous conditions materialize: a substantial increase in base layer on-chain activity, a revitalization of decentralized finance, and greater adoption of real-world applications (RWA, tokenization, payments). The note is not a short-term bearish price target, but a thesis of relative underperformance against BTC. The bank points to the 95% year-over-year collapse in L1 fees, the ETH/BTC ratio at a ten-month low (0.02835), and the slow recovery of flows in spot ETH ETFs compared to BTC ETFs as concrete evidence.
Why do L2 rollups harm the price of ETH if Ethereum remains the base layer?
Because they break value capture without breaking technical utility. After EIP-4844, rollups pay minimal fees to publish data on L1: in 2024, they paid about $113 million (41% of their revenue); in 2025, barely $10 million, a 90% year-over-year drop. This deactivates the EIP-1559 burning mechanism: if base fees tend to zero, net burning also tends to zero, and consensus issuance comes to dominate. The result is a net growth in ETH supply of 0.23% annually, which breaks the ultrasound money narrative that supported the asset against BTC in previous cycles.
What on-chain metrics should be monitored to validate or refute the banking thesis?
Four metrics summarize the debate. First, median L1 fees (Etherscan): currently 0.5 gwei after a 95% year-over-year drop. Second, blob usage on L2: the more demand there is, the closer the data market gets to a price floor that will burn ETH again. Third, DeFi TVL on mainnet (DefiLlama): dominance has fallen from 63.5% at the beginning of 2025 to 54% in May 2026, with $45.4 billion locked on L1. Fourth, net flows to spot ETH ETFs (ETHA, ETHE, ETHB): currently recovering only one-third of previous outflows compared to two-thirds for BTC. Without a reversal in these four metrics, the JPMorgan thesis holds.
Do the upcoming Glamsterdam and Hegota upgrades solve the value capture problem?
They solve the technical problem, not necessarily the economic one. Glamsterdam (Q3 2026) quintuples the L1 gas limit to 200 million, enables parallel processing (BALs, EIP-7928), and brings proposer-builder separation into the protocol (ePBS, EIP-7732), which is attractive for institutional compliance. Hegota (Q4 2026) introduces Verkle Trees, reduces state weight by 90%, and brings native smart accounts (EIP-8141). But if the increased capacity is not filled with new organic demand, the only thing that happens is further compression of the base fee and, with it, the burn. JPMorgan insists on this paradox: three years of upgrades have moved in the direction of reducing costs, not creating demand.
What do traditional banks miss when analyzing Ethereum with this framework?
Three things. First, that value capture for a base layer can migrate to staking (implicit yield) and re-staked services (EigenLayer, AVS) without going through the EIP-1559 burn: 30% of the supply is staked, and ETFs with staking (Grayscale, ETHB) drain ETH from free float. Second, that RWA and tokenized money market funds (BlackRock's BUIDL, already over $2.9 billion on Ethereum) consolidate the network as a regulatory settlement base, not a speculative casino. And third, that banking accounting frameworks still value ETH as a software stock (DCF on fees) when it is more akin to a monetary asset with demand for economic security.
When will JPMorgan's next note confirming or revising this thesis be published?
JPMorgan does not communicate a fixed schedule for crypto notes. Panigirtzoglou's team typically publishes revisions every 4-8 weeks when there is a material change in institutional flows (ETFs, CME futures, hedge fund positioning). It is reasonable to expect a new note after the next milestones: monthly close of May and June for ETF flows, quarterly TVL data from DefiLlama at the end of Q2, and especially after the activation of Glamsterdam in Q3 2026. In the meantime, interim reports on Bitcoin dominance and CTA positioning often contain lateral references to Ethereum that allow for triangulating the bank's position without waiting for a dedicated note.