Coinbase Cut 700 Citing AI. The Ex-Coinbase Rust Pool Is ~125.

Coinbase announced on Tuesday, May 5, 2026 that it will cut roughly 700 people, about 14% of staff, and CEO Brian Armstrong framed the reduction as an AI-native restructuring rather than a market retreat. If you recruit protocol engineers, the interesting question is not the 700. It is how many of them can actually ship Rust against an L2 codebase, and whether you can find them before every restaking startup does.

Short answer: the defensible Rust and protocol slice is around 100 to 125 people, not the several thousand your LinkedIn "past company: Coinbase" filter will return.

What Coinbase actually announced on May 5, 2026

Coinbase is cutting ~700 employees (14% of headcount), citing market conditions and AI changing how the company operates. The layoffs will finish by Q2 2026, will cost $50M to $60M in restructuring charges, and leave the company with about 4,300 people.

Three details matter for anyone sourcing the fallout:

• Armstrong told staff the company will have "no pure managers" and will compress to a maximum of five layers between executives and ICs.
• Coinbase is standing up "AI-native pods," including one-person teams that direct agents doing the work of engineers, designers, and PMs.
• Departing US staff get at least 16 weeks of base pay, extra weeks by tenure, health coverage support, and their next equity vesting.

That severance package is the single most important variable in your outreach strategy. These engineers can wait 60 to 90 days. Any message that leads with urgency will lose to a message that respects their runway.

4,300

Why "ex-Coinbase" on LinkedIn is a garbage filter

The "past company: Coinbase" filter returns thousands of alumni across three crypto winters, and the noise-to-signal ratio for protocol engineering is roughly 10 to 1. Filtering by title makes it worse, not better.

Coinbase has been through this before:

• June 2022: ~1,100 laid off (18%)
• January 2023: ~950 laid off (another ~20%)
• May 2026: ~700 laid off (14%)

Stack those cohorts and you get a multi-vintage alumni universe dominated by support, compliance, marketing, business ops, and engineers whose last production commit was to a repo that no longer exists. A naive Boolean like "ex-Coinbase" AND ("Rust" OR "Solidity") will over-return by an order of magnitude, because a lot of those keywords sit in "About" blurbs from people who audited one contract in 2021.

The signal you actually want is narrower: left Coinbase in the last 90 days, with commit history on Base, Rosetta, or the MPC library. That is closer to a GitHub query than a LinkedIn one, which is the exact gap Refolk closes: describe the person in plain English, and I rank across GitHub, LinkedIn, and the open web instead of pretending job titles are ground truth.

The real ex-Coinbase Rust pool, sized honestly

The defensible estimate is 100 to 125 hireable Rust and protocol ICs from the May 2026 cut, out of roughly 280 to 315 engineers total. Here is the math.

Coinbase does not publish its engineering ratio, but exchanges typically run 40% to 45% of headcount in engineering. Apply the 14% cut proportionally against ~5,000 people and you get 280 to 315 engineers exiting. Of those, the Rust and protocol slice (Base, Rosetta, MPC, wallet, keystore) plausibly sits at 30% to 40% of Coinbase's engineering org given its public Base and OP Stack investment. That lands at ~100 to 125 ICs with real Rust and onchain chops.

That is the number to plan around. Not 700. Not "thousands of ex-Coinbase engineers on LinkedIn." Roughly 125. Treat it as a modeling estimate, not a measured headcount - the exact figure will only settle as WARN filings and profile updates land.

The 2,357 figure is the one that should scare you into moving fast. That is the entire US ceiling of engineers who list both Rust and Solidity as skills. If even 100 ex-Coinbase Rust ICs hit the market in the same 30-day window, that is a ~4% one-week supply shock to the national onchain-Rust pool. Every L2, wallet, restaking, and MPC custody startup will notice at the same time.

2,357

Where the good ones actually work now

The top current employers of US Rust+Solidity engineers are not other exchanges. They are infra, DeFi, and dev-tools startups that quietly absorbed the last two Coinbase waves.

In Refolk's index, the top current employers of the US Rust+Solidity cohort are:

• Oxide Computer Company (3 in the sample, ~12% share)
• Figure
• Mintlify
• Udemy
• Cozy Finance
• Credit Coop

Two implications. First, most of the strongest Rust+Solidity engineers are already placed at companies where they are happy, technically challenged, and paid in a mix of cash and equity that is not trivial to beat. Second, the ex-Coinbase May 2026 cohort will not compete against other exchanges for their next role. They will compete against Oxide, Figure, and a stack of infra Series B companies you may not currently be tracking.

Geography narrows the search further. Per Refolk's index sample, the US Rust+Solidity pool concentrates in four metros:

• San Francisco: 4 of 25 (~16%)
• NYC + Brooklyn: 4 of 25 combined
• Austin: 2 of 25
• Everywhere else: long tail

If your sourcing plan includes Denver, Seattle, Miami, or remote-first outreach into secondary metros, you are burning cycles. Concentrate spend where the pool actually lives.

Every L2, wallet, and restaking startup is about to fish the same pond in the same 30 days.

The AI-native framing is a filter for you, not against you

The May 2026 cut skews toward two very different groups: AI resisters and mid-level coordinators whose work is being absorbed by agents. The hireable slice is the second group. Armstrong has said publicly, on John Collison's podcast in September, that Coinbase has fired employees for not using AI tools, which means the surviving org is already AI-tooled and the exits are structurally selected.

Break the cohort into three buckets:

1. AI resisters. Engineers who declined to adopt Cursor, Claude Code, or internal Coinbase AI tooling. Skip these unless your stack is deliberately low-AI.
2. Pure managers and mid-level coordinators. Coinbase is compressing to five layers and eliminating "pure managers," pushing span of control past the Gallup 2024 average of 12.1 reports. Some of these people were ICs two years ago and can code again. Many cannot. Interview carefully.
3. Senior ICs who lost their team, not their skills. This is the group you want. They are AI-tooled, they shipped protocol code, and they were caught in a structural reorganization, not a performance cull.

The third bucket is where the ~100 to 125 Rust and protocol number lives. Refolk lets you write that filter as a sentence ("senior Rust engineers who left Coinbase in the last 90 days with commits on Base or Rosetta") instead of stacking LinkedIn filters that were never designed to answer the question.

Base is the honeypot, not Coinbase

The highest-leverage Rust talent Coinbase employed did not have "Coinbase" in the most useful part of their profile. It had "Base," "OP Stack," or "Rosetta" in their GitHub commit history.

Base is Coinbase's L2, joined as the second Core Dev team on Optimism's OP Stack. The engineers who worked on it are, functionally, OP Stack Core Devs with a Coinbase paycheck. That distinction matters because:

• OP Stack contributors are visible on GitHub, not LinkedIn. Their most valuable signal is a merged PR against ethereum-optimism/optimism or base-org/*, not a job title.
• Rosetta maintainers (Coinbase's blockchain integration spec) are similarly identifiable through coinbase/rosetta-* repos.
• The open-sourced Coinbase MPC cryptography library has a small, identifiable contributor list. Every one of those contributors is a premium hire for wallet, custody, and account-abstraction startups.

The right sourcing move for this cohort is a GitHub-first search, cross-referenced against recent employment change signals. That is the workflow Refolk is built for: I read the commit graph, the profile, and the open web at the same time, and I return the humans, not a stack of filter chips.

A 30-day sourcing plan for the ex-Coinbase Rust pool

The window is short and the pool is small, so run in parallel, not in sequence. Here is the sequence that works.

1. Days 1 to 3: Define the ~125. Build a GitHub-first list of commit authors against base-org/*, ethereum-optimism/optimism, coinbase/rosetta-*, and the Coinbase MPC library, filtered to people whose Coinbase tenure ended after May 5, 2026.
2. Days 4 to 10: First-touch outreach. Lead with the technical work, not the layoff. Reference a specific PR. Do not use the word "urgently." These engineers have 16+ weeks of severance and will filter out any message that pattern-matches to panic.
3. Days 11 to 20: Second-touch and warm intros. Route through their Oxide, Figure, or Base-adjacent connections. Warm intros from ex-Coinbase engineers already at your peer companies will out-convert cold email 3 to 5x on this cohort.
4. Days 21 to 30: Onsite and offer. Compress your loop. Two rounds, one system design, one code review of a Rust PR they are proud of. Ex-Coinbase seniors have already survived one 8-round loop this year. Do not make them do it again.

The startups that will land this cohort are the ones treating it as a 30-day named list, not an inbound funnel. If you are still filtering LinkedIn by "Rust" 60 days from now, the pool will already be at Oxide and Figure.

FAQ

How many ex-Coinbase engineers with Rust skills actually hit the market in May 2026?

Roughly 100 to 125, out of an estimated 280 to 315 total engineers cut. That estimate applies a typical 40 to 45% engineering ratio to Coinbase's ~5,000 headcount, then takes the 30 to 40% Rust and protocol slice implied by Coinbase's public work on Base, Rosetta, and the MPC library. The exact number will move as WARN filings and LinkedIn updates land, but the order of magnitude is low hundreds, not thousands.

Why doesn't a LinkedIn "past company: Coinbase" search work for this?

Because Coinbase has laid off staff in three separate waves (2022, 2023, 2026) totaling roughly 2,750 people across every function, so the alumni universe is multi-vintage and heavily non-engineering. The noise-to-signal ratio for protocol engineering on that filter is roughly 10 to 1. The useful signal is a recent departure date combined with commit history against Base, Rosetta, or the MPC repo, which is a GitHub query, not a LinkedIn one.

How urgent is the sourcing window if severance is 16+ weeks?

The engineers are not urgent, but the market is. With only 2,357 US engineers listing both Rust and Solidity in Refolk's index, even 100 new candidates is a ~4% supply shock, and every L2, wallet, and restaking startup will notice within the same 30 days. Move quickly on identification and outreach, but do not project urgency in your messaging. The candidates have runway; your competitors do not.

Should I also be sourcing from other AI-cited layoffs at the same time?

Yes. The recruiters winning this cohort are also watching Gemini Space Station, Block, Pinterest, CrowdStrike, and Chegg, all of which cited AI in recent cuts. The skill overlap with Coinbase is highest at Gemini, where the exchange and custody stack is closest. Treat the AI-cited layoff wave as a single sourcing surface, not a series of one-off news events, and run a persistent named-list query rather than reacting to each headline.