---
title: "Crypto-Native Firms"
book: "The Industry: Firms, Roles and Careers"
subject: quant
language: en
chapter: 10
exercises: 8
source: https://one-course.com/books/quant/17/en/chapter/10-crypto-native-firms
---

# Chapter 10 — Crypto-Native Firms

The largest US-listed crypto exchange reported 4 510 employees at the end of 2022, after a restructuring in June that had cut about 18% of its staff; in January 2023 it cut about 21% more, and a year later it had 3 416. Its revenue had fallen by 59% in 2022. In 2024 its revenue more than doubled, and by the end of 2025 it employed 4 951 people. In crypto, headcount follows revenue, which follows the market for the assets the industry trades, with a lag of about a year; and many employees are paid in those assets as well. This chapter describes crypto-native firms as employers: what kinds there are, what is different about working in them, how their headcount moves with the cycle, what pay in tokens is worth, and what happens to an employee when such a firm fails.

## 10.1 Kinds of employer

Crypto-native firms are those whose business began in crypto assets rather than moving into them.

- **Market makers and trading firms** quote on centralised exchanges and trade on chain (Book 11, chapters 24–25), and make markets for token issuers under agreements (Book 3, chapter 24).
- **Funds** invest in tokens and in the companies and protocols that issue them.
- **Exchanges and brokers** run centralised venues (Book 3, chapter 15), custody and on-ramps between money and crypto assets.
- **Protocol teams and foundations** build and fund the software of a blockchain or an application on it.

**Definition 10.1 (Protocol foundation).**

A *protocol foundation* is a non-profit organisation that funds and coordinates the development of an open-source blockchain protocol and its ecosystem, typically from a treasury held partly in the protocol’s own token, and that does not own the protocol it supports.

One foundation describes itself as “a non-profit that supports the Ethereum ecosystem” and adds that it is “not a tech company, or a normal non-profit”. Its treasury policy shows what funding from a token means for the people it pays: “periodically selling ETH to ensure sufficient savings for future years, and programmatically increasing our fiat savings in bull markets to fund spending in bear markets”. A foundation’s budget, and so its hiring, depends on the price of what it holds.

## 10.2 What is different: round-the-clock operations, custody, token pay, regulatory status

Four things set crypto-native employers apart from the rest of the industry.

1. **The markets never close** (Book 3, chapter 29, calls the practice follow-the-sun trading): on-call rotas, weekend incidents and global teams are the norm, not the exception (chapter 26).
2. **Custody is the firm’s own problem.** Keys, wallets and the operational security around them are a core engineering job (Book 3, chapter 24), and a failure is a loss, not an outage.
3. **Pay is often partly in tokens** , whose value moves with the market and whose tax treatment differs by country (section 4).
4. **Regulatory status is still being settled** in many jurisdictions (Book 3, chapter 24, on the EU’s Markets in Crypto-Assets Regulation): firms move, license, and restructure as rules arrive, and employees move with them.

## 10.3 Headcount through the cycle

**As of December 2025 — One listed crypto exchange through the cycle.**

Coinbase (Forms 10-K for 2022 to 2025): total revenue $7 839 million (2021), $3 194 million (2022), $3 108 million (2023), $6 564 million (2024), $7 181 million (2025); employees at year end 4 510 (2022), 3 416 (2023), 3 772 (2024), 4 951 (2025). In June 2022 a restructuring affected about 18% of headcount (about 1 100 employees); in January 2023 a second affected about 21% of the end-2022 headcount (about 950 employees).

The pattern is lagged ([Figure 10.1](#fig-in-crypto-native-firms-cycle)). Revenue fell 59% in 2022 and 3% in 2023; headcount fell 24% in 2023, the year after the revenue collapse. Revenue rose 111% in 2024 and 9% in 2025; headcount rose 10% in 2024 and 31% in 2025. A firm hires on last year’s revenue and cuts on it too, so an employee joining at the top of a cycle joins the cohort most exposed to the next cut. Five years of one firm is not a law, but it is the pattern a crypto job seeker should expect until the evidence says otherwise.

![One listed crypto exchange, 2021–2025: total revenue and year-end employees. Employees for 2021 are not in the filings read. Headcount follows revenue with about a year’s lag. Data: Coinbase Forms 10-K for 2022–2025, through in_crypto.headcount.](https://one-course.com/images/onecourse/chapters/quant-17/in-crypto-native-firms/fig-9e68517ec8ad.svg)

***Figure 10.1.** One listed crypto exchange, 2021–2025: total revenue and year-end employees. Employees for 2021 are not in the filings read. Headcount follows revenue with about a year’s lag. Data: Coinbase Forms 10-K for 2022–2025, through `in_crypto.headcount`.*

Revenue per employee shows the same cycle from the employee’s side ([Figure 10.2](#fig-in-crypto-native-firms-perhead)): $0.71 million per year-end employee in 2022, $0.91 million in 2023, $1.74 million in 2024 and $1.45 million in 2025, a swing by a factor of 2.5 in two years. Chapter 12 measured the same sensitivity across firms: crypto revenue per head moved against the equity market’s volatility, not with it. A bonus pool set from revenue moves with this line; a headcount set from last year’s revenue lags it, which is when cuts come.

![One listed crypto exchange’s total revenue per year-end employee, 2022–2025. Data: Coinbase Forms 10-K, through in_crypto.per_head.](https://one-course.com/images/onecourse/chapters/quant-17/in-crypto-native-firms/fig-225901722ee5.svg)

***Figure 10.2.** One listed crypto exchange’s total revenue per year-end employee, 2022–2025. Data: Coinbase Forms 10-K, through `in_crypto.per_head`.*

## 10.4 Paid in tokens

**Definition 10.2 (Token compensation, token warrant).**

*Token compensation* is pay delivered in a crypto asset, usually under a vesting schedule with a cliff, and often with a lock-up after vesting during which vested tokens may not be sold. A *token warrant* is a right, granted with equity in a company, to receive tokens that the company or an affiliate may issue in the future, in proportion to the equity held.

Tax authorities treat tokens received for work as pay. The US tax authority’s guidance states that virtual currency paid by an employer as remuneration for services constitutes wages for employment tax purposes; the UK’s states that cryptoassets received as employment income count as money’s worth and are subject to income tax and National Insurance on their value. The value taxed is the value at receipt, usually at vesting. If the token then falls during a lock-up, the employee has paid tax, in cash, on value that no longer exists.

**Method 10.3 (Valuing a token grant).**

1. Write the schedule: the cliff tranche, the monthly tranches, the lock-up after each vesting.
2. Simulate the token’s price in the grant’s currency over the vesting period and the lock-up, with a stated volatility and drift (parameters, not forecasts).
3. For each tranche, compute the tax due at vesting and the proceeds at the first date it may be sold, net of any liquidity discount.
4. Report the distribution of net proceeds against the same grant value paid in cash, and how often a tranche’s tax exceeds its proceeds.

**Example 10.4 (A four-year grant).**

A grant worth $400 000 at the token’s price on the grant date vests 25% after a one-year cliff and then monthly over three years; each tranche is locked up for six months; tax is 40% of its value at vesting (illustrative). With a volatility of 80% a year and no drift, the median net outcome over 20 000 simulated paths is $106 193, against $240 000 for the same grant paid in cash; the 10th and 90th percentiles are $17 467 and $537 810; the mean, $235 062, is close to the cash figure. In 76.2% of paths at least one tranche’s tax exceeds what it later sells for ([Figure 10.3](#fig-in-crypto-native-firms-grant)). At a volatility of 40% the median is $194 318 and such a tranche occurs in 2.4% of paths.

![The net outcome of a $400 000 four-year token grant (25% cliff at one year, monthly thereafter, six-month lock-up, 40% tax at vesting), 20 000 simulated price paths at 80% annual volatility and no drift; outcomes above $1 million are counted in the last bin. The dashed line is the same grant paid in cash. Illustrative parameters. Data: firm.tokencomp.simulate, through in_crypto.grant_hist.](https://one-course.com/images/onecourse/chapters/quant-17/in-crypto-native-firms/fig-4de37812aaf7.svg)

***Figure 10.3.** The net outcome of a $400 000 four-year token grant (25% cliff at one year, monthly thereafter, six-month lock-up, 40% tax at vesting), 20 000 simulated price paths at 80% annual volatility and no drift; outcomes above $1 million are counted in the last bin. The dashed line is the same grant paid in cash. Illustrative parameters. Data: `firm.tokencomp.simulate`, through `in_crypto.grant_hist`.*

The distribution is skewed: most paths leave the employee with less than cash would have, a minority with much more. The mean is close to the cash value because the price has no drift, which is exactly why the median is not: at 80% volatility the typical path loses value. A candidate comparing a token grant with cash is comparing a lottery of about equal expectation with a certainty, before counting the tax that falls on value at vesting.

## 10.5 When the firm fails: the employee in the bankruptcy record

Crypto firms can fail quickly (Book 3, chapter 15, and Book 16, chapter 29, describe a large exchange failure). For an employee, a failure turns pay into a claim. In a US bankruptcy, unpaid wages, salaries and commissions earned in the 180 days before the petition have priority among unsecured claims only up to a fixed amount per individual, $17 150 since 1 April 2025; the rest ranks with other unsecured creditors. Unvested tokens and equity in the failed firm are usually worth nothing. The practical lessons are those of any young firm, sharper: take cash where possible, sell vested tokens on the schedule you planned, and know which entity employs you.

## 10.6 Tutorial: headcount and a token grant

**Goal.** Read one crypto exchange’s headcount against its revenue, and value a token grant. **End state:** Figures [10.1](#fig-in-crypto-native-firms-cycle) and [10.3](#fig-in-crypto-native-firms-grant).

1. **The filings.** `data/industry/crypto_headcount.csv` holds year-end employees and total revenue from the exchange’s Forms 10-K; `in_crypto.changes()` computes the yearly changes.
2. **The grant.** `firm.tokencomp.Grant(400_000, months=48, cliff=12, lockup=6, tax=0.40)` and `simulate(grant, vol, drift, n, rng)` ([Listing 10.1](#lst-in-crypto-native-firms-sim)). `def simulate (g, vol, drift, n, rng): horizon = g.months + g.lockup dt = 1 / 12 z = rng.standard_normal((n, horizon)) logp = np.cumsum((drift - 0.5 * vol ** 2 ) * dt + vol * math.sqrt(dt) * z, axis=1 ) price = np.exp(np.concatenate([np.zeros((n, 1 )), logp], axis=1 )) # price relative to grant date, month 0..horizon net = np.zeros(n) gross = np.zeros(n) tax = np.zeros(n) exceeds = np.zeros(n) for m, frac in schedule(g): at_vest = g.value0 * frac * price[:, m] at_sale = g.value0 * frac * price[:, m + g.lockup] * (1 - g.discount) t = g.tax * at_vest gross += at_sale tax += t exceeds += t > at_sale net += at_sale - t cash = g.value0 * (1 - g.tax) return dict (net=net, gross=gross, tax=tax, cash=cash, tax_exceeds=exceeds / len (schedule(g)))` **Listing 10.1.** Each tranche taxed at vesting and sold after its lock-up, over simulated price paths. code/firm/tokencomp/firm_tokencomp.py
3. **Summarise.** `firm.tokencomp.summary` gives the median, the 10th and 90th percentiles, the mean, the cash equivalent and the share of paths with a tranche whose tax exceeds its proceeds.

**What to change next.** Lengthen the lock-up to twelve months and see how often tax exceeds proceeds (exercise 7); give the token a positive drift and ask what drift makes the median equal to the cash value.

## 10.7 Build: the token-grant model

**Purpose.** Put a number, and a distribution, on pay in tokens, for this chapter and for chapter 13’s comparison of pay packages.

**Interface.** `firm.tokencomp`: `Grant(value0, months, cliff, lockup, tax, discount)`; `schedule(grant)`; `simulate(grant, vol, drift, n, rng)`; `summary(sim)`.

**Rules.** Tax is due at vesting on the value then; a tranche is sold at the end of its lock-up; the price model’s drift and volatility are the caller’s stated assumptions; the schedule sums to the grant.

**Acceptance tests.** `code/firm/tokencomp/tests/`: the schedule sums to one with a quarter at the cliff; at zero volatility the outcome equals the cash value net of tax; volatility spreads the outcome, lowers the median below the cash value and keeps the mean near it.

**Stretch.** Jumps in the price (Book 4); a tax regime that taxes at sale; hedging a grant with futures or options where they exist and are allowed.

Sources and further reading

- Coinbase Global, Inc., Forms 10-K for 2022, 2023, 2024 and 2025.
- US Internal Revenue Service, Notice 2014-21; HM Revenue & Customs, Cryptoassets Manual, CRYPTO21100.
- Ethereum Foundation, home page and EF Report 2024.
- US Code, title 11, section 507, and the Judicial Conference’s adjustment of dollar amounts effective 1 April 2025.

## 10.8 Exercises

**Exercise 10.1 ★.**

Compute the exchange’s change in revenue and in headcount for each year from 2022 to 2025.

**Solution of Exercise 10.1.**

Revenue: $-59.3\%$ (2022), $-2.7\%$ (2023), $+111.2\%$ (2024), $+9.4\%$ (2025). Headcount: $-24.3\%$ (2023), $+10.4\%$ (2024), $+31.3\%$ (2025); 2021’s headcount is not in the filings read.

**Exercise 10.2 ★.**

In the example, how much of the grant vests at the cliff, and how much each month after?

**Solution of Exercise 10.2.**

25% ($100 000 at the grant price) at the cliff, then $1/48$ of the grant ($8 333) each month for 36 months.

**Exercise 10.3 ★.**

An employee receives tokens worth $10 000 at vesting, taxed at 40%, and sells them after a lock-up at half the price. What is the net outcome?

**Solution of Exercise 10.3.**

Tax $0.4\times10\,000=\$4\,000$; proceeds $5 000; net $1 000.

**Exercise 10.4 ★★.**

Why is the median outcome of the grant below the cash value while the mean is close to it?

**Solution of Exercise 10.4.**

With no drift the expected price stays constant, but the lognormal median falls as $e^{-\sigma^2t/2}$: a few paths rise a lot and many fall. The mean is carried by the rare large outcomes; the typical outcome is below it.

**Exercise 10.5 ★★.**

A failed firm owes an employee $60 000 of wages earned in the last four months. How much has priority under the US rule?

**Solution of Exercise 10.5.**

$17 150; the remaining $42 850 ranks as a general unsecured claim.

**Exercise 10.6 ★★.**

From [Figure 10.1](#fig-in-crypto-native-firms-cycle), in which two years did headcount change most out of step with revenue, and what explains both?

**Solution of Exercise 10.6.**

2023 (headcount $-24\%$ as revenue fell only 3%) and 2025 (headcount $+31\%$ as revenue rose 9%): in both, headcount was responding to the previous year’s revenue, the collapse of 2022 and the doubling of 2024.

**Exercise 10.7 ★★★.**

*Coding.* Rerun the grant with a twelve-month lock-up. How do the median and the share of paths in which a tranche’s tax exceeds its proceeds change?

**Solution of Exercise 10.7.**

The median falls from $106 193 to $76 275, and the share of paths with a tranche whose tax exceeds its proceeds rises from 76.2% to 86.8%: a longer lock-up leaves more time for the price to fall after tax is fixed.

**Exercise 10.8 ★★★.**

*Find the flaw.* “The token grant’s expected value equals the cash offer, so the two offers are equivalent.”

**Solution of Exercise 10.8.**

Equal expectation is not equivalence: the grant’s median is less than half the cash value, its outcomes range from almost nothing to more than twice cash, and tax falls on value at vesting whatever happens later. A risk-averse employee values the grant below its mean (chapter 13’s certainty equivalent), and the firm’s own survival is correlated with the token.

## 10.9 Problem: Paid in Tokens

**Problem 10.1.**

Weekend problem — paid in tokens

An engineer is offered a salary plus a four-year token grant worth $400 000 at today’s price, or the same salary plus $400 000 in cash over four years. She wants to understand the difference.

**Part I — The employers.**

1. Name the four kinds of crypto-native employer.
2. Define a [protocol foundation](#def-in-crypto-native-firms-foundation) and describe how one foundation funds itself through the cycle.
3. Give four ways crypto-native employers differ from the rest of the industry.
4. Give the exchange’s revenue and headcount from 2021 to 2025.
5. What were the two restructurings, in share and number of employees?

**Part II — Tax.**

6. Define [token compensation](#def-in-crypto-native-firms-tokens) and a [token warrant](#def-in-crypto-native-firms-tokens) .
7. How do the US and UK tax authorities treat tokens received for work?
8. When is the tax due, and on what value?
9. What happens if the token falls during the lock-up?
10. Compute the net outcome of a $10 000 tranche taxed at 40% and sold at half its vesting price.

**Part III — The grant.**

11. Write the vesting schedule.
12. State the method for valuing the grant.
13. Give the median, the 10th and 90th percentiles and the mean at 80% volatility.
14. What is the cash equivalent, and in what share of paths is the grant worth less?
15. In what share of paths does at least one tranche’s tax exceed its proceeds, at 80% and at 40% volatility?

**Part IV — The verdict.**

16. State the *named result* : the median and the 10th–90th percentile range of the grant at 80% volatility, against cash, and the share of paths in which tax exceeds proceeds.
17. How much of her unpaid wages would have priority if the firm failed?
18. What does the headcount history suggest about when to join?
19. Name two ways to reduce the grant’s risk, if allowed.
20. In two sentences, how should she choose?

**Solution of Problem 10.1.**

1. Market makers and trading firms, funds, exchanges and brokers, protocol teams and foundations.
2. A non-profit that funds and coordinates an open-source protocol from a treasury partly in its token; one sells its token periodically and builds fiat savings in bull markets to fund spending in bear markets.
3. Round-the-clock markets; self-custody; pay in tokens; unsettled regulatory status.
4. Revenue $7.8, 3.2, 3.1, 6.6, 7.2 billion; employees (2022–2025) 4 510, 3 416, 3 772, 4 951.
5. June 2022, about 18% (about 1 100); January 2023, about 21% of end-2022 headcount (about 950).
6. Pay in a crypto asset under a vesting schedule, often with a lock-up; a right attached to equity to receive future tokens.
7. As wages (US) and as employment income subject to income tax and National Insurance (UK).
8. Usually at vesting, on the value then.
9. The employee has paid tax on value that has gone.
10. $1 000.
11. 25% at twelve months, then $1/48$ monthly to month 48; each tranche locked six months.
12. Schedule; simulate prices; tax at vesting and proceeds after lock-up per tranche; distribution against cash.
13. $106 193; $17 467 and $537 810; $235 062.
14. $240 000 ($400 000 less 40% tax); in 74.3% of paths the grant is worth less.
15. 76.2% at 80% volatility; 2.4% at 40%.
16. Median $106 193, 10th–90th percentile $17 467 to $537 810 against $240 000 in cash; 76.2% of paths with a tranche taxed on more than it sells for.
17. Up to $17 150.
18. Joining at the top of a cycle puts her in the cohort hired on peak revenue and most exposed to the next cut.
19. Sell tranches as soon as allowed; hedge with futures or options on the token where they exist and her contract allows; negotiate cash for part of the grant or tax at sale.
20. The cash offer is worth more to her unless she wants a lottery on the token and can bear losing most of it; if she takes the grant, she should plan to sell on schedule and budget the tax in cash.

## 10.10 Interview questions

**Interview question 10.1 ★ developer.**

What does a crypto exchange’s engineering team do that a stock exchange’s does not?

**Solution of Interview question 10.1.**

Custody of client assets (key management, wallets), on-chain deposits and withdrawals with confirmations, round-the-clock operation without a closing auction, and security against attackers who can steal irreversibly.

*What the interviewer is looking for: custody and irreversibility.*

**Interview question 10.2 ★ trader.**

Why do crypto trading firms staff round the clock, and how would you organise a desk to do it?

**Solution of Interview question 10.2.**

The markets never close and move most when others sleep; a follow-the-sun desk hands the book between regions, with clear handover of positions, limits and open incidents, and an on-call engineer per shift.

*What the interviewer is looking for: handover discipline.*

**Interview question 10.3 ★★ researcher, risk.**

A token’s price is lognormal with 80% volatility and no drift. What is the median of its price in one year, relative to today?

**Solution of Interview question 10.3.**

$e^{-\sigma^2/2}=e^{-0.32}=0.726$ of today’s price.

*What the interviewer is looking for: the lognormal median.*

**Interview question 10.4 ★★ risk.**

You are paid in a token you cannot sell for a year. How could you hedge, and what would stop you?

**Solution of Interview question 10.4.**

Short futures or perpetuals on the token, or buy puts, if they exist and are liquid; limits: the employment contract or insider rules may forbid it, the market may be too thin, and funding costs may be high.

*What the interviewer is looking for: instruments, and the contractual and practical constraints.*

**Interview question 10.5 ★★ developer, risk.**

A firm keeps most client assets in cold wallets and some in hot wallets. What decides the split, and what can go wrong?

**Solution of Interview question 10.5.**

Expected withdrawal flow and its tail against the risk of holding keys online: enough in hot wallets to meet withdrawals, the rest cold. Risks: theft from hot wallets, delays and operational errors in moving from cold storage.

*What the interviewer is looking for: liquidity against security.*

**Interview question 10.6 ★★★ researcher.**

With five years of one firm’s revenue and headcount, how would you test whether headcount lags revenue, and what would you conclude?

**Solution of Interview question 10.6.**

Correlate headcount changes with revenue changes at lags zero and one: here the lag-one relation is the stronger. With four or five points, report it as a pattern, not a test, and look for other firms’ data before generalising.

*What the interviewer is looking for: lags, and humility with five points.*