L2s Hit 100k+ TPS But We're Still Paying L1 Gas—When Does Scaling Actually Feel Cheap?

[ethereum_emma]

I just paid $45 in L1 gas to bridge some ETH from mainnet to Arbitrum. For a protocol that’s supposed to “scale Ethereum,” this doesn’t feel very scaled.

Here’s the thing that’s been bothering me: Ethereum L2s collectively process 100,000+ transactions per second now—that’s 6,500x mainnet’s ~15 TPS capacity. Arbitrum and Optimism are doing 2-4k TPS each, zkEVMs are targeting 10k+ TPS, and we’re seeing specialized L2s for privacy, gaming, and ultra-low latency applications.

L2 transactions themselves are cheap—we’re talking $0.05 to $0.50 per transaction. That part works beautifully. But the moment you need to bridge assets to/from L1, you’re back to paying mainnet gas fees. And with current gas prices, that can be $30-$80 for a single bridge operation.

The L2 Scaling Paradox

On L2: Transactions are blazing fast and incredibly cheap
Bridging to/L2: Slow (7 days for withdrawals on optimistic rollups) and expensive (L1 gas costs)

The technical reality is that L2s still depend on L1 for:

• Security: L2s inherit Ethereum’s validator set security
• Settlement: Transaction data gets posted to L1 for verification
• Bridging: Moving assets between layers requires L1 transactions

For optimistic rollups specifically, the 7-day withdrawal window isn’t a bug—it’s a security feature. The network needs time for validators to submit fraud proofs if something goes wrong. ZK-rollups improve on this with validity proofs, but they still need to post proofs to L1.

When Does “Scaling” Actually Feel Like Scaling?

From a user perspective, here’s the problem:

• If I’m making 100+ transactions on an L2, the bridge cost amortizes well
• But for casual users making 10-20 transactions, the economics break down
• You’re paying $45 to access a system where transactions cost $0.25

Some questions for the community:

1. Should L2 protocols subsidize bridging costs for users? (Maybe cover first bridge for new users?)
2. Is the “stay on L2, minimize bridging” advice realistic for most users?
3. Are we solving scaling by creating a two-tier system—power users who live on L2s, casual users who can’t justify the bridge cost?
4. Will native L2-to-L2 messaging/shared sequencers solve this, or is high L1 bridging cost just the permanent price of security?

I’ve been working on L2 infrastructure for years, and I genuinely believe this tech is game-changing. But I also think we need to be honest about the UX gaps that still exist. Telling users “L2s are cheap!” and then hitting them with a $45 bridge fee creates a trust problem.

What’s your experience been? Are you staying on L2s long-term, or does the bridging cost still make you think twice?

[blockchain_brian]

You’re hitting on something that frustrates a lot of people, but there’s a fundamental trade-off here that I think is worth understanding.

L1 gas isn’t a bug—it’s the cost of security.

When you bridge to an L2, you’re essentially paying the Ethereum validator set to provide security guarantees for your assets. That $45 isn’t arbitrary—it’s what it costs to get your transaction included in a block that’s secured by ~1 million validators and billions of dollars in staked ETH.

L2s are cheap precisely because they batch hundreds or thousands of transactions together and only pay L1 gas once for the entire batch. You’re getting the benefit of that batching when you transact on the L2. But someone has to pay for the initial entry and exit—and that’s you, the user doing the bridging.

The 7-Day Window Is Security, Not Laziness

For optimistic rollups (Arbitrum, Optimism, Base), that 7-day withdrawal window isn’t a technical limitation we can just “fix”—it’s the security model working as designed. The network needs time for anyone to submit fraud proofs if the sequencer posts invalid state transitions.

zkRollups are improving this (instant finality once the validity proof is verified on L1), but they still need to post proofs to mainnet, which costs L1 gas.

Practical Solutions That Exist Today

1. Stay on L2 longer: If you’re making 100+ transactions, bridging costs amortize to basically nothing per transaction. The UX pain is mostly for people who bridge in, make 5 trades, and bridge out.

2. Native L2-to-L2 messaging: This is coming. Projects like OP Stack’s Superchain and shared sequencer networks (Espresso, Astria) will allow you to move between L2s without touching L1 at all.

3. CEX direct withdrawals to L2: Coinbase, Binance, and others now let you withdraw directly to Arbitrum/Optimism/Base. This bypasses the L1 bridge entirely for most users.

4. Third-party bridges with liquidity pools: Projects like Hop Protocol and Across use liquidity pools so you can “instant withdraw” without waiting 7 days. You pay a small fee to LPs, but it’s often cheaper than L1 gas if you’re in a hurry.

Is This The New Normal?

Probably yes, but it’s also getting better. As L2 adoption grows:

• More users will onboard directly to L2s (never touching L1)
• L2-to-L2 communication becomes seamless
• CEXs make L2 withdrawals the default

The vision isn’t “everyone bridges all the time”—it’s “most people live on L2s and rarely need L1 at all.”

But yeah, if you’re a power user who needs to move between L1 and L2 frequently, you’re going to pay for that security. That’s the trade-off.

[data_engineer_mike]

This conversation got me curious, so I pulled some data on recent L1→L2 bridge transactions. Here’s what I found analyzing 10,000 bridge operations from the past month:

Bridge Cost Breakdown (March 2026)

L1 → L2 Bridge Costs:

• Median: $28
• Average: $34
• Range: $12 (off-peak) to $87 (peak congestion)
• 75th percentile: $45

L2 Native Transaction Costs:

• Median: $0.12
• Average: $0.18
• Range: $0.03 to $0.85

So Emma’s $45 bridge cost is pretty typical, maybe slightly high-end but not unusual.

Break-Even Analysis

Here’s the math on when bridging makes sense:

Scenario 1: Casual User (10 transactions)

• Bridge cost: $45
• L2 transactions: 10 × $0.15 = $1.50
• Total: $46.50
• Break-even vs. L1: ~2 transactions

Scenario 2: Active User (100 transactions)

• Bridge cost: $45
• L2 transactions: 100 × $0.15 = $15
• Total: $60
• Per-transaction average: $0.60

Scenario 3: Power User (1000+ transactions)

• Bridge cost: $45
• L2 transactions: 1000 × $0.15 = $150
• Total: $195
• Per-transaction average: $0.195

The economics work great if you’re staying on the L2. But for someone making 10-20 transactions, you’re paying $2-4 per effective transaction when you factor in bridging. That’s still cheaper than L1, but it’s not the “$0.15 per transaction” headline people see.

The Two-Tier System Is Real

Looking at wallet behavior patterns:

• Top 5% of L2 users: Average 200+ transactions/month, bridge 1-2x/year
• Bottom 50% of L2 users: Average 8 transactions/month, bridge 3-4x/year

Power users bridge rarely and stay on L2 = amazing economics.
Casual users bridge frequently and make few transactions = still paying significant costs per transaction.

Brian’s totally right that “stay on L2 longer” solves this, but it requires a mindset shift. Most people are used to CEX → do a few trades → CEX, not “live permanently on L2.”

What Actually Needs to Change

IMO the real solution is what Brian mentioned: CEX direct L2 onboarding. If Coinbase/Binance make L2 deposits/withdrawals the default, most users never touch L1 at all. They go fiat → CEX → L2 without ever paying a bridge fee.

That’s already happening (Coinbase Base withdrawals, Binance Arbitrum support), but it needs to become the default, not an option buried in settings.

The other thing that would help: L2 protocols subsidizing the first bridge for new users. If you’re trying to attract users, eating that $45 cost for their first deposit is probably worth it to get them into the ecosystem.

[layer2_lisa]

Really appreciate these responses—this is exactly the kind of discussion I was hoping for.

@blockchain_brian You’re absolutely right that the 7-day window is a security feature, not a bug. I should have been clearer in my original post. My issue isn’t with the technical design—optimistic rollups work as intended. It’s the UX expectation mismatch.

When we market L2s as “instant, cheap transactions,” new users hear that and expect the entire experience to be instant and cheap. Then they hit the bridge and it’s neither instant nor cheap. That’s where the trust problem comes in.

Your point about zkRollups is spot on though. We’re seeing Starknet, zkSync, and Polygon zkEVM offer much faster finality because validity proofs don’t require the fraud proof window. But even there, proof generation takes time and costs L1 gas to submit. It’s better, not perfect.

Mike’s Data Is Gold

@data_engineer_mike This breakdown is fantastic. The two-tier system you identified is the real issue:

• Power users (5%) get amazing economics because they amortize bridge costs
• Casual users (50%) still pay $2-4 per effective transaction

This creates an adoption barrier. If you want mass adoption, you can’t tell casual users “just make 100+ transactions per month to justify the bridge cost.” That’s not how most people use crypto.

Your point about CEX direct L2 onboarding becoming the default is exactly where this needs to go. Coinbase already does this well with Base—you can deposit USD and withdraw directly to Base without ever touching L1. That should be the standard pattern for every L2.

What I’d Love to See from L2 Protocols

1. First-bridge subsidies: Arbitrum, Optimism, zkSync—if you want new users, subsidize their first $50 bridge. That $45 they save gets them 200-300 L2 transactions. Enough to fall in love with the platform.

2. Better UX around L2-to-L2 transfers: Projects like Across, Hop, and Bungee exist but most users don’t know about them. L2 protocols should integrate these directly into their bridge UIs with clear cost comparisons.

3. Transparent cost expectations: Stop marketing “L2 transactions cost $0.15” without the asterisk “but bridging costs $30-80.” Show total cost of entry upfront.

Is High Bridge Cost Permanent?

Brian mentioned OP Stack’s Superchain and shared sequencers (Espresso, Astria). These are game-changers if they work at scale. Native L2-to-L2 messaging means you could move between Optimism, Base, and other OP Stack chains without touching L1 at all.

If that works, and CEXs default to L2 withdrawals, most users will never pay bridge fees. They’ll onboard via CEX → L2, live on L2, and move between L2s using shared infrastructure.

At that point, L1 bridging becomes a “power user” feature for people who specifically need mainnet exposure. And if you’re a power user moving large amounts, $45 is probably acceptable.

The Vision vs. Today

The vision: Users live on L2s, rarely touch L1, move seamlessly between chains.

Today: Users still bridge frequently, pay significant L1 gas, and experience the pain of 7-day withdrawal windows.

We’re making progress, but we’re not there yet. And I think it’s important to be honest about that gap rather than pretending the problem doesn’t exist.

Thanks for the thoughtful discussion!