Ethereum ETF outflows contradict bullish staking and layer-2 metrics

Ethereum’s institutional data hub reported roughly $120 billion in staked ETH and $40.4 billion in daily average total value locked across layer-2 networks in a Sept. 21 snapshot, but the two figures measure separate things and cannot be summed to gauge fresh demand for the asset. Institutional investors watching the numbers should note that US-traded spot Ethereum ETFs still posted over $140 million in net outflows from September 15 to 18, showing large on-chain balances can coexist with capital leaving a specific investment channel.

• Staked ETH stood near $120 billion versus $40.4 billion in daily average L2 total value locked as of Sept. 21.
• US spot Ethereum ETFs bled $405.4 million from Sept. 14 to 17 before closing the week with $143.7 million in inflows on Sept. 18.
• Ultrasound.money showed a 1.8 gwei execution base fee on Sept. 21, a level that limits how much ETH is burned per unit of gas.

• $120B total staked ETH securing the network as of Sept. 21
• $40.4B daily average TVL across layer-2 networks the same day
• $140M net ETF outflows recorded between Sept. 15 and 18

Ethereum’s headline metrics look large in isolation, but staking data and L2 asset totals track different flows entirely. One reflects ETH locked into validators for network security, the other reflects capital sitting on scaling networks that settle back to the base chain.

Neither figure, on its own or combined, answers the question institutional allocators actually care about: how much new ETH is being bought.

Staking Balances Conflate Old Holdings With New Purchases

Staking works by depositing ETH to activate validators that secure the network and earn rewards, a mechanism detailed on ethereum.org. An investor can stake ETH they already own, which raises the staking total without representing any new buying. They can also acquire ETH specifically to stake it, but the result is not necessarily fresh demand in all cases, and the dollar-denominated staking figure cannot distinguish between the two paths.

Liquid staking tokens complicate the picture further. They let holders transfer exposure to staked ETH while the underlying deposit stays committed to validation, meaning the token’s presence in a wallet says nothing about when or why the original ETH was acquired.

Farside Data Shows ETF Flows Reversed Twice in a Week

Ethereum ETF flows offer a cleaner demand signal, but only for the slice of capital that moves through those funds. Farside Investors’ Ethereum ETF table shows the week opened with $121.1 million in inflows on Sept. 14, then reversed into $405.4 million of outflows through Sept. 17, before US spot funds closed the week on Sept. 18 with $143.7 million back in the door.

Net outflows across Sept. 15 to 18 still totaled more than $140 million despite that late rebound. The swing illustrates how quickly ETF sentiment can flip within a single trading week even as the broader staking and L2 ecosystem barely moves.

Farside’s data captures only ETF-channel activity, so it cannot resolve whether total ETH demand rose or fell over the same period.

Gas Fees at 1.8 Gwei Separate L2 Activity From ETH Burn

The $40.4 billion sitting on L2 networks measures assets held there, not payments those networks make to Ethereum. L2BEAT’s on-chain-costs measure tracks what L2 operators actually pay for posting calldata, blobs, compute, and overhead, and that spending is distinct from the fees users pay an L2 directly.

Even total operator spending does not equal ETH burned. Ethereum’s execution base fee is burned while priority fees go to validators, and blob fees operate in a separate market that is also burned.

Ultrasound.money showed a 1.8 gwei gas header on Sept. 21, a low reading that means less ETH is burned per unit of gas consumed than at higher fee levels. That can reflect either weaker network demand or successful scaling that made transactions cheaper, and total burn still depends on gas consumed alongside issuance to determine Ethereum’s net supply change.

The CCS read. We think the real story is that institutional dashboards are reporting stock, not flow, and allocators pricing ETH off staking or TVL headlines are measuring the wrong variable. The ETF ledger, however volatile week to week, remains the cleanest proxy available for actual purchasing decisions, which is exactly why its swings deserve more weight than aggregate balance figures.

Farside Investors updates its Ethereum ETF table daily, making the next print of net flows the most immediate gauge of whether last week’s Sept. 18 rebound holds or whether outflows resume; separately, any sustained rise in gas fees above the 1.8 gwei level tracked by Ultrasound.money on Sept. 21 would signal renewed burn pressure worth watching against issuance.