Will it pay?

What one regulator costs to make

The detailed accounting is in What it cost to engineer a marine alternator regulator1. This analysis needs some figures reconfigured the way a filed Cost of Sales is so that the margins compare apples to apples to other companies.

With production cost known, the next question is what to sell a regulator for.

Routes to the consumer

An important realization for me, having been consistently force fed with  "buy direct and save!" and "available now at dealer cost!" type advertising: direct selling does not magically remove the distributor/dealer channel's cut for most cases. It just replaces it with costs the manufacturer then must carry itself: creating demand (advertising, content, shows), pre-sale advice and quoting, order handling, packing and outbound shipping, card fees, returns, first-line support, warranty handling, and carrying inventory.  Unless a manufacturer can do these more efficiently than the available dealers (or reduce the need), it's not good business.  The intermediaries' margin generally pays for work somebody has to do, itemized player by player in the appendix.

Now some working channel assumptions for the rest of this article:

• the dealer takes 30% of the shelf price
• the distributor takes 15% of its own selling price, which is 40.5% of retail together
• courting a dealer network costs the manufacturer 3.2% of its revenue, Garmin's advertising line2
• servicing a channel account costs another 7.3%, Dragonfly's OEM segment cost line8

If real contract terms turn out different, the arithmetic gets rerun then; section 10 shows how much the answer moves if they do.

What each route does to price, and X Engineering's share

The same regulator, through each route, with demand ignored: if every route sold the same number of units, what shelf price and what manufacturer economics does each one create? The profit column is per unit before any engineering cost, and on the direct rows it already carries selling and marketing at 20% of revenue, an assumption explained in its own section below.

Pros and Cons of each Route

The $655 in rows D and E is what holding the $390 trade price does to a shelf price once a distributor takes 15% and a dealer takes 30%. Priced the other way, off Garmin Marine's 55.0% gross margin2, the same regulator lands at $624. The $690 in row B was set without reference to either, and sits 5% above the first and 11% above the second.

Marketing runs at two rates through this table: 20% of revenue where X Engineering has to create consumer demand itself9, and 3.2% where a dealer creates it and X Engineering only courts and supplies the dealer.

Rows A and B, or any price between them, are the only choices I can pick from today, and choosing is largely a bet on volume. $390 has to move 3.04 units for every one $690 moves in order to bring in the same profit, so cutting the price 43% has to buy more than 3X the units. 

Against row D, the dealer route priced for the chain, the ratio is gentler: 1.57 units for every one.

Units a year to earn $50,000

$50,000 is a random round target. Operating profit before engineering expenses.

Both curves are steep on the left and flat on the right. The first $150 of price addition (over the promotional price) buys most of the relief, and past about $650 more price barely moves the unit count to earn a fixed profit.

The gross margins say the same thing from another direction. Hold the shelf price at $390 and hand the unit to a chain, and the dealer's 30% and the distributor's 15% come out of my end instead of the customer's: I receive $232, and against $164.29 of cost of sales that is a 29.2% gross margin, under Navico Group's 36.0% and Johnson Outdoors Fishing's 30.4%, and barely half Garmin Marine's 55.0%4,3,2. Those margins are struck before engineering on both sides: under GAAP, research and development is expensed below the gross profit line, so Garmin Marine's 55.0% is calculated before the 15.9% it spends on R&D, the same way mine is calculated before the 4,314 hours. The same unit sold direct leaves 57.1%, above all three, and that gap is the whole prize in the direct model. To earn what Garmin Marine earns through a dealer chain the regulator would have to retail at $624, which is near where Wakespeed's $649 WS500 already sits11.

How big is the market?

Registration counts times lithium-adoption guesses give almost any answer, so the useful question is whether anyone has sold thousands of these things a year, and luckily Wakespeed has.

Dragonfly Energy bought Wakespeed in April 2022 with an earnout, and its filings disclose that $3.0M of Wakespeed-branded and Wakespeed-IP product cleared within 20 months, deemed earned by the end of 20238. That is $1.8M a year. At the WS500 Pro's $649 list price11 it is about 2,800 units a year, and at a $400 wholesale price, 38% below list (these are my estimates), it is 4,500. Call it 2,800 to 4,500 regulator-equivalent units a year. Arco is the other data point: well over a thousand Zeus regulators in the field about six months after shipping12.

The Wakespeed number shows the market supports several thousand units a year from one brand. What X Engineering will sell is a separate question: Wakespeed got there with a decade of dealer relationships, an installed base, and a name that installers already specify. For a first product from a brand nobody has heard of, Wakespeed scale is the long-run target. So the volume cases used from here on are 500, 1,000, and 1,500 units a year, with 3,500 a year, a round number near the middle of the Wakespeed range, as the upside case.

Two forces pull against each other on the size of the market. ABYC E-13 has required charge sources to follow the battery manufacturer's profile since July 202313, which rules out a stock alternator charging a lithium bank directly and leaves two compliant answers on a retrofit, an external regulator or a DC-DC charger (a much less powerful option), and registrations of boats 26 feet and over grew 4.1% while the whole fleet grew 1.1%14. Against that, new boat retail fell 8.8% in 2025, accessory spending slipped from $12.4B to $12.1B15, and Dragonfly's own direct sales fell while it moved toward OEM. It seems a growing niche inside a shrinking market.

What raising the price does to volume

The generic elasticity literature cannot say what XREG-010's real demand curve looks like, but we can look at sensitivity.

Is the price and volume curve knowable in a market like this one? Claude says:

This structure argues for low price elasticity. A regulator is bought roughly once per boat per decade, and rare purchases get decided on not-being-wrong rather than on price. The part exists to protect a $ alternator, $$ lithium bank, and $$$ boat. Installation labor frequently exceeds the part's price, so a $600 difference shrinks as a share of the project.

Being better and cheaper without a track record is its own problem. Nelson's search-good argument18 says verifiable specifications substitute for advertising, and your comparison matrix is exactly that. But Milgrom and Roberts19 and the Zahavi handicap argument20 run the other way: visible spend and a high price are costly signals that a firm expects repeat business. A product claiming to beat a $1,050 incumbent10 at 63% less cost asks a stranger to believe something that normally isn't true. Some fraction of buyers will read the price as evidence against the specification sheet.

The binding constraint isn't elasticity, it's the consideration set. Price elasticity only operates among options a buyer is already weighing. For a brand nobody has heard of, cutting price doesn't recruit buyers, it lowers the margin on the ones who would have bought anyway.

Profit vs. Price

The (meaningless) anchor is 500 units a year at today's $390, so every curve starts at the same $53,680.

Direct route only. Dots mark where each curve peaks. The dashed line is today's profit at 500 units and $390.

At a (very unlikely) elasticity of −3.0, the peak is $364, below today's price, and worth 1% more than today. At −2.5 it is $405 and worth almost nothing. At −2.0 it is $485 and worth 6%. At −1.5 it is $728 and worth 29%.

The elasticity at which $450 earns exactly what $390 earns is −2.39, at $550 it is −2.14, and at $690 it is −1.95. All three sit inside the range the published work reports, between Tellis's −1.76 and Bijmolt's −2.62, so the literature does not settle this. The answer depends on which end of that range an alternator regulator sits at.

At $450, the worst of the four elasticities in the chart costs 8% of annual profit and the best gains 14%. At $550 the spread runs from losing 26% to gaining 25%, and at $690 from losing 45% to gaining 29%. The downside grows faster than the upside at every step, and only the $450 move keeps the loss in single digits. So $450 carries little modeled downside, $550 materially more, and $690 is a serious bet on how price sensitive buyers are.

$390 is the right price for the first 50 boards, to get them into the field and find out what the support burden and failure modes are, but it is not a price the business can live on. The realistic direct price is between $550 and $690, and the top of that range is about where a dealer has shown interest. The real elasticity has to be learned by changing the price and watching, which in a market this small will be a noisy measurement.

What it costs to create a sale

There is no published customer-acquisition cost for a new marine alternator regulator brand either, so this is another unknown. The available starting point is what comparable technical hardware businesses spend to sell, and the working assumption throughout this article is 20% of revenue for marketing, selling and support, taken from the selling-and-marketing line at seven technical hardware makers, SolarEdge at 9.9% up to Digi International at 21.3%, with the ones that sell a real share "direct" at the top of that band9. It is a fully loaded selling-and-marketing benchmark, broader than a customer acquisition cost: it includes advertising, sales activity, trade shows, content, and the rest of the commercial machinery. For a brand-new hardware company where every sale is to a newly acquired customer it is a fair first-order estimate, maybe on the low end, of the commercial burden of generating sales.

At $390 that is $78 a sale. At $690 it is $138. Nothing guarantees that the cost of finding a buyer scales with the price, but absent real data a percentage of revenue is a better approximation than a fixed dollar figure carried across very different prices, and there is no point polishing the model further before there is anything to calibrate it against. Section 10 tests it at 10% and 30%.

$78 of every $390 sale is higher than I would have guessed (as an engineer learning about this stuff as I go), and arguably it should be higher still for a new product which both lacks market credibility and benefits from network effects. That argues for sacrificing short term profit and getting as many out there as possible. But that's directly contrary to the generic advice given to small niche businesses, which always says to keep initial price high, go for the premium segment, sell less, keep support burden low, and then later expand and try to compete on price. It's complicated stuff, impossible for me to predict what marketing strategy and price will maximize short and long term viability of the business.

What can be bounded is the most I can afford to pay to make one sale. Each additional sale costs $204.64 in cash and brings in $390, so $185.36 is the contribution, before anything is spent on finding the buyer or on supporting them afterward, and it is the budget every marketing decision comes out of. Three ceilings on it, from loosest to tightest:

The profit-maximizing rule is the one an accountant would state: keep spending on demand while the next dollar still returns more than a dollar of contribution, and stop when the marginal sale costs $185. In practice nobody can measure the marginal sale that precisely, so the working rules are the three ceilings above, and the 20% assumption sits between the tightest and the middle one.

What the channel's $155 buys, at the same $390 shelf price, is shelf space, a salesperson the customer already trusts, and an installer who specifies the part. Matching that reach for $78 a unit without them is the hardest part of the "direct" plan, and the one likely to decide if this works.

So far, discovery is through cheap or free methods.  Forum threads, videos, Facebook, cheap google ads, word of mouth.  Support comes from the developer, me, and the operating assumption is that it stays that way. Routine support is meant to be handled by Claude, with the documentation and the software written so that self-service works better than any product of this sort in history. Substantial human engineering involvement could be a separate service at $150 an hour, billed by the minute, rather than an unlimited free obligation, because huge support burden is definitely another way this all could fail. I have no plan or desire to hire employees, so the step change in payroll that a conventional small hardware company eventually hits has to be designed out. This is the model being tested on the first units, and support economics get their own article once there is data to write it from.

What the business earns in a year

Price times units, less the production cost from the first section, less 20% for creating the sale, less the freight, card fee, warranty, scrap, and admin already listed. Per unit on the direct route that is $107.36 at $390, $151.10 at $450, $224.00 at $550, and $326.06 at $690; through a distributor and dealer at a $655 shelf it is $169.02. Run across the volume cases:

Operating profit a year before any engineering. Revenue is units times price on the direct columns; on the chain column X Engineering collects $390 of the $655.

At 500 units and $390 the business earns $53,700 a year, and the same 500 units at $550 earn twice that. Replacing a $150,000 salary (sr. level EE/developer), loaded at 1.43x for what an employer pays21, takes 1,998 units a year at $390, 958 at $550, and 658 at $690. Garmin Marine runs engineering at 15.9% of revenue2; at that ratio, $400,000 a year of engineering is the budget that goes with $2.5M of revenue, which is 6,400 units at $390 or 3,600 at $690.

Was the engineering worth it?

None of the pricing above tries to recover the hours already spent on XREG-010. They are sunk, and most of them were fun, education, and AI experimentation that would have happened anyway. The price is chosen on the forward economics of the business, and this section asks a separate question: given that business, how long would its profits take to economically repay the development investment?

The tracked figure is 4,314 hours1. At $50 to $150 an hour, the span from a junior rate to what a contract electrical engineer bills22, that is $216,000 to $647,000. The round case used here is 4,000 hours at $100 an hour, $400,000, and the table is years to repay it out of the annual figures above:

Years to repay $400,000 of engineering. Multiply by 0.5 for $50 an hour and by 1.5 for $150.

In units rather than years, the $400,000 comes back after 3,726 sold at $390, 2,647 at $450, 1,786 at $550, or 1,227 at $690. At $390 and 500 units a year the hours are never really repaid, since seven and a half years is longer than the product will stay current. At $550 and 1,000 units a year they come back in under two, and at Wakespeed scale in about a year at any price. Every unit after that is profit before new engineering.

What price and route to choose

The pricing question: how much consumer surplus can X Engineering leave with the buyer while keeping enough producer surplus to stay alive, carry inventory, survive warranty failures, keep improving products, and pay for life at least as well as a corporate job does?

Even before writing, I felt more at home selling direct vs. using distributors, dealers, and OEM's, probably because I've been the consumer, and I can fully wrap my head around expectations. But this research also showed additional reasons.  As priced, it's too cheap, it's not going to be profitable enough for a dealer to justify vs. more expensive incumbents that do essentially the same job, so I don't have a pitch to them, especially before the product is proven. One exception might be "installing dealers", who don't depend as much on hardware markup because they're billing labor, which is not subject to OEM price rules and is therefore likely much more profitable. Here, I could see XREG-010 getting a foot in the door with unusual applications that aren't worth support from the market leaders. Flexibility, direct access to the developer (me), and speed are likely to be advantages for X Engineering in these early days.

As a new entrant, X Engineering has nothing to protect, no dealer relationships to upset and no shelf space to lose, so it's cheap to experiment and iterate. The opening I plan: direct sales, an unusual share of the money going into engineering, and demand created by the product's measured performance and word of mouth rather than by 20% of the price spent on marketing.

Decisions:

Targets that follow from the tables, as a plan: the first 50 units at $390 through the end of 2026, then 500 units at $550 in 2027, which is $112,000 of operating profit before engineering and a little under a third of the $400,000 repaid. In 2028, 1,000 units at $550, which is $224,000 and leaves $64,000 of the engineering to come back early in 2029; at $690 the same 1,000 units repay all of it. Wakespeed scale, 3,500 units a year, is the long-run competitive target, and at $690 it is about where Garmin's engineering ratio would fund $400,000 a year of new engineering.

What could change the answer

Appendix

A. The players in the chain, and what each one is paid for

Who sits between the factory and the boat, what each of them wants, and what the margin they take pays for. Background to the channel assumptions in section 2.

This is "coopetition" with multiple players around each company.  Customers, suppliers, competitors, and complementors, with the same player often sitting in multiple boxes at once. A player's bargaining power in the network is its added value, meaning what the whole thing would lose if the player left, so the way to earn a better split is to get harder to replace25.

Double marginalization, from Spengler26, says that when two independent players each take a margin the consumer price ends up higher and the combined profit lower than if one firm priced the chain end to end. Holding retail fixed does not escape that. It fixes the consumer's price and pushes the whole dealer-and-distributor cut onto the manufacturer. A channel is also a repeated game rather than a single trade, and Axelrod's tournaments27 found that the strategies which win repeated games start cooperative, retaliate immediately, and forgive quickly, which is why underselling dealers pays once and then costs for years.

The same mutual dependence that stabilizes the traditional channel also discourages economic disruption. A manufacturer can innovate technically, and Garmin, Furuno, and Victron definitely do. But radically changing price, distribution, support, or margin structure threatens the economics of every other participant in the chain: distributor margin, dealer margin, pricing parity, minimum advertised prices, existing accounts, inventory, and support expectations all depend on each other. So the business model stays put, theoretically leaving opportunity for a more efficient approach.

Where a manufacturer feeds several routes at once, the highest shelf price is applied to all of them, and minimum list and advertised prices exist to hold that line, keeping parity for all players28. At any one shelf price the long channel route pays the manufacturer least, rows D and E being the clearest case at $169.02 against $300.55, and manufacturers accept that gap for what the channel adds in other value.

B. Rows A and C of the route table, line by line

The arithmetic behind the profit column in section 3, for the same $390 shelf price through the two routes.

Take admin at 2.2% off both columns and 20% selling off the direct column and the last row becomes the $34.29 and $107.36 in the route table.

Sources

1. X Engineering blog, "What it cost to engineer a marine alternator regulator" (product cost per unit, shipping, card fee, tracked hours): https://xengineering.net/blog/what-the-regulator-cost/
2. Garmin Ltd, FY2025 Form 10-K (filed 2026-02-18), Marine segment, channel sentence, direct-channel share, risk factors: https://www.sec.gov/Archives/edgar/data/1121788/000119312526056028/grmn-20251227.htm
3. Johnson Outdoors Inc, FY2025 Form 10-K (filed 2025-12-12), Fishing segment, channel list, customer concentration, marketing and admin lines: https://www.sec.gov/Archives/edgar/data/788329/000114036125045348/jout-20251003.htm
4. Brunswick Corporation, FY2025 Form 10-K (filed 2026-02-13), Navico Group segment, product lines, distribution businesses and dealer count: https://www.sec.gov/Archives/edgar/data/14930/000001493026000027/bcorp-20251231.htm
5. Garmin Ltd, FY2021 Form 10-K, Amazon customer disclosure: https://www.sec.gov/Archives/edgar/data/1121788/000095017022001303/grmn-20211225.htm
6. Boating Industry, 2026-05-18, West Marine files for Chapter 11, revenue and creditors: https://boatingindustry.com/breaking-news/2026/05/18/west-marine-files-for-chapter-11-bankruptcy/
7. West Marine Chapter 11 case site (Verita), Case No. 26-10794, D. Del.: https://www.veritaglobal.net/westmarine
8. Dragonfly Energy Holdings Corp, FY2025 Form 10-K (filed 2026-03-30), direct and OEM segments and their cost lines, warranty accrual: https://www.sec.gov/Archives/edgar/data/1847986/000149315226013635/form10-k.htm ; FY2023 10-K with the Wakespeed earn-out: https://www.sec.gov/Archives/edgar/data/1847986/000149315224014831/form10-k.htm
9. Selling and marketing as a share of revenue, latest annual report on Form 10-K for each: SolarEdge Technologies, Enphase Energy, Power Integrations, GoPro, Sonos, iRobot, Digi International, and Vicor. Retrieved from SEC company facts, 2026-09-12.
10. X Engineering, Alternator Regulator Product Comparison Matrix, 1 September 2026, complete apples-to-apples prices: https://xengineering.net/blog/alternator-regulator-comparison/
11. Wakespeed WS500 Pro list price $649, PKYS Inc. (pkys.com) listing, retrieved September 2026
12. IBEX 365 interview with Arco Marine, Zeus regulators in the field, 2024-05-28: https://www.ibexshow.com/ibex365/innovation-awards/2023-ibex-innovation-award-interview-with-arco-marine/
13. Panbo, ABYC ratifies E-13, the first lithium battery standard, 2022-08-08: https://panbo.com/abyc-ratifies-e-13-their-first-lithium-battery-standard/
14. U.S. Coast Guard, Recreational Boating Statistics 2024, Table 37, registered recreational vessels by length: https://www.uscgboating.org/library/accident-statistics/Recreational-Boating-Statistics-2024.pdf
15. National Marine Manufacturers Association, 2025 retail statistics and accessory spending: https://www.nmma.org/press/article/25432
16. Gerard J. Tellis, "The Price Elasticity of Selective Demand: A Meta-Analysis of Econometric Models of Sales", Journal of Marketing Research 25(4), 1988, 331-341: https://journals.sagepub.com/doi/10.1177/002224378802500401
17. Tammo H. A. Bijmolt, Harald J. van Heerde and Rik G. M. Pieters, "New Empirical Generalizations on the Determinants of Price Elasticity", Journal of Marketing Research 42(2), 2005, 141-156: https://journals.sagepub.com/doi/10.1509/jmkr.42.2.141.62296
18. Phillip Nelson, "Information and Consumer Behavior", Journal of Political Economy 78(2), 1970, and "Advertising as Information", Journal of Political Economy 82(4), 1974.
19. Paul Milgrom and John Roberts, "Price and Advertising Signals of Product Quality", Journal of Political Economy 94(4), 1986.
20. Amotz Zahavi, "Mate Selection: A Selection for a Handicap", Journal of Theoretical Biology 53(1), 1975.
21. U.S. Bureau of Labor Statistics, Employer Costs for Employee Compensation, June 2026: https://www.bls.gov/news.release/ecec.nr0.htm
22. U.S. Bureau of Labor Statistics, Occupational Outlook Handbook, Electrical and Electronics Engineers, May 2025 wages: https://www.bls.gov/ooh/architecture-and-engineering/electrical-and-electronics-engineers.htm
23. The Coast Distribution System, FY2014 Form 10-K, gross margin and SG&A: https://www.sec.gov/Archives/edgar/data/728303/000119312515113703/d867960d10k.htm
24. Coast Distribution FY2010 Form 10-K: https://www.sec.gov/Archives/edgar/data/728303/000119312511084885/d10k.htm
25. Adam M. Brandenburger and Barry J. Nalebuff, Co-opetition, Doubleday, 1996: the Value Net and added value.
26. Joseph J. Spengler, "Vertical Integration and Antitrust Policy", Journal of Political Economy 58(4), 1950: double marginalization.
27. Robert Axelrod, The Evolution of Cooperation, Basic Books, 1984: repeated-game tournaments.
28. Furuno USA Authorized Reseller Policies and MRP List, May 2019 (dealer-hosted copy): https://cedmarine.com/wp-content/uploads/2020/08/FURUNO-Authorized_Reseller_Policies_and_Application-2019.pdf