Django Software Foundation revenue was $296,584 in 2024, and 84% of the spending left the United States

Quick answer (September 2026). The Django Software Foundation reported total revenue of $296,584 in 2024, $316,832 in 2023 and $267,843 in 2022 on its Form 990. Expenses were $277,682, $275,792 and $285,330, so 2024 closed with a surplus of $18,902 and net assets of $274,497. Those are small numbers for the foundation behind one of the most widely deployed web frameworks in the world. The more interesting number is not in Part I at all. It is on Schedule F, where the DSF reports $234,085 of 2024 expenditure booked to activities outside the United States, which is 84.30% of everything it spent. A North Carolina charity registered in Kansas spends roughly five dollars in six abroad, and it does it with zero employees.

The 2024 return is the one almost nobody has read. It was signed on 13 November 2025 and it arrived in the IRS 2025 index, which is late enough that the main free summary of this organisation still has not picked it up in the place most tools look.

The filings this is built from

Two Form 990s, pulled from the IRS filed-XML archives rather than from any summary, covering the years ended 31 December 2023 and 31 December 2024. The DSF's EIN is 26-2708549. The 2024 return carries object id 202533179349310693 and sits in the 2025 index; the 2023 return is 202403189349311825 in the 2024 index. Figures for 2022 come from the prior-year columns of the 2023 return and from the public summary record.

The IRS index for 2025 is a plain CSV, and the filings behind it are the only form in which a whole return is machine readable, schedules included. One practical note for anyone repeating this: the 2025 archives are compressed with deflate64, which the Python standard library refuses to open. The 2024 archives are ordinary deflate. Getting the 2024 return out required writing an inflater by hand and checking the result against the CRC-32 the archive itself stores, which matched at 0x57a6abd7 across 36,215 bytes.

Both returns are prepared on the cash basis and neither is audited. The 2024 return answers "no" to the question of whether the financial statements were compiled, reviewed or audited by an independent accountant. That matters for what follows, because it means there is no second set of figures to reconcile the return against. The only other financial account the DSF publishes is its own annual report, and that turns out to be a useful check in its own right.

What the 2024 return says

Revenue fell 6.39% in 2024, and contributions fell 7.37%, from $312,492 to $289,456. The surplus survived because expenses barely moved, rising $1,890 on the year.

Revenue is almost entirely donated money:

The net assets roll forward exactly in both years: $214,555 plus the 2023 surplus of $41,040 gives $255,595, and $255,595 plus the 2024 surplus of $18,902 gives $274,497. There are no prior period adjustments and no unrealised gains to reconcile. The balance sheet is about as simple as a balance sheet gets. Total assets at the end of 2024 were $274,548 and total liabilities were $51.

Where the money actually goes

Part IX of the 2024 return has eight non-zero lines, and three of them are grants. Collapsed into six categories, that is the entire cost base of the Django Software Foundation.

The share column is rounded to two decimals and adds to 100.01, for the same reason the DSF's own pie chart adds to 100.1 further down. Six rounded percentages do not have to sum to a round number, and neither case is an error.

The single line at the top is the Fellowship. The return's own programme description puts $218,134 against the Django Fellowship, the arrangement under which a small number of paid people triage tickets, review patches and keep the release process moving. The other two programme lines are local event grants at $27,650 and support for Django Girls at $10,000, and the three add to $255,784, which is exactly the reported programme services total.

Management and general was $21,898, and its four components add to that figure exactly. Fundraising expense was $0, on $289,456 of contributions raised. The return also answers "no" to whether the organisation conducted any fundraising activities.

The Fellowship money is booked as fees for services rather than as wages, and the reason is on Part V. The DSF reports zero employees. Not a small number, zero. Part I line 5 and Part V line 2a both read 0 for 2024 and for 2023.

The 84% that leaves the country

Schedule F is where a $300,000 charity stops looking like a $300,000 charity.

Against total expenses of $277,682, the 2024 figure is 84.30%. The 2023 figure was 89.03% of that year's $275,792. Both years, the overwhelming majority of the DSF's spending is booked to activities outside the United States, and the trend is a slow shift from Europe toward South America rather than any move back home.

This is not a grants story. Of that $234,085, only $23,950 is grants to foreign organisations and individuals. The rest is the Fellowship. Schedule F's regional table counts three people in column (d) for 2024, two in Europe and one in South America, and that column asks for employees, agents and independent contractors together, which is why it reads three while the main form correctly reports zero employees. The two numbers do not contradict each other. They are counting different things, and reading the column heading is what separates a finding from a mistake.

What it does mean is that the maintenance of Python's most widely used web framework is, financially, an export. American companies fund it, and the work is done elsewhere.

The reserve is the number to watch, and it is fine

For an infrastructure non-profit, revenue is a poor health metric. The reserve is better: net assets divided by annual expenses, expressed in days of cover.

Just under a year of cover, and improving for two consecutive years. That is worth saying plainly because this series has recently covered organisations where the same calculation was alarming. When Ruby Central's reserve is run through the identical formula it bottoms out at 11.5 days. The DSF is at 360.8. On the single measure that most often predicts trouble for a volunteer-run foundation, the Django Software Foundation is in good shape.

Two caveats belong with that. Net assets are 100% without donor restriction in both years, with the donor-restricted element simply absent from the returns, so the reserve really is spendable. And the entire balance sheet sits on the line marked cash, non-interest-bearing: $274,548 there, $0 in savings and temporary cash investments, $0 in publicly traded securities. The same return reports $6,912 of investment income, which is an implied 2.608% on the average balance for the year. Money that earns 2.6% is not sitting in a non-interest-bearing account. Either the classification on line 1 is doing work the label does not describe, or the income arrived from somewhere the balance sheet does not show. The return does not settle it, so it is recorded here as an observation rather than a conclusion.

The annual report and the return tell the same story in different words

The DSF published its 2024 Annual Impact Report on 30 June 2025, announced on its own weblog that day. The financial paragraph reads, in full: "For our 2024 budget of USD $296,583; $195,300 of our revenue comes from Corporate Donors, and $94,156 comes from individual donors. Expenses comprise $229,893 of wages, $38,172 of grants, $9,536 of professional fees including hosting. We also generated a surplus of $18,982."

Set against the return, that reconciles better than it first looks.

Three of those map exactly. The two donor categories sum to the filed contributions figure to the dollar. The report's "wages" is the Fellowship fees plus salaries, to the dollar. The report's "grants" is the filed grants line plus the $522 of occupancy, to the dollar.

The report also foots against itself: $296,583 less $277,601 of expenses is $18,982, exactly the surplus it states. So the small gaps against the return, $1 on revenue and $81 on expenses, are not internal errors in the report. They are the difference between two accounts of the same year, and the whole $81 sits in the professional fees bucket.

The likeliest reason is simply the calendar. The report was published on 30 June 2025. The return was signed on 13 November 2025, 136 days later. The report's figures came first and the return's came after the books closed. That is an unremarkable explanation, and it is worth stating because the alternative reading, that someone got the numbers wrong, is the one a reader reaches for first and it is not supported.

One thing in that paragraph is worth flagging anyway. The figure labelled "our 2024 budget" is within $1 of the actual filed revenue for the year, and its two components equal the filed contributions figure exactly. It is not a budget. It is the year's income, described with the wrong noun, in the only financial paragraph the organisation publishes.

The report also carries a pie chart headed "Grants around the world in 2023-2024": Global 38.4%, Africa 24.8%, North America 18.6%, Europe 12.4%, Oceania 3.6%, Asia 2.0%, South America 0.3%. Those sum to 100.1%, which for seven categories rounded to one decimal place is ordinary rounding and not an error. The more useful caveat is what the chart covers. Grants are 13.56% of DSF spending. The chart is an accurate picture of about one dollar in seven, and the other six are the Fellowship, whose geography is the Schedule F table above and looks nothing like the pie.

Three things that do not reconcile

Everything checked so far foots. These three do not, and they are small, specific and checkable.

A $2,000 gap in foreign grants. The 2024 Part IX reports $25,950 of grants to foreign organisations, governments and individuals. Schedule F itemises $10,000 and $6,000 to two organisations in Europe and $7,950 across 14 individuals in East Asia and Pacific, which is $23,950. The $2,000 difference is not explained on the return. The same test on the 2023 return reconciles exactly: $6,000 plus $10,000 plus $8,362 across 18 individuals gives $24,362, matching Part IX to the dollar. A test that passes exactly one year and fails by $2,000 the next is an anomaly, not a reporting convention.

A recipient's EIN, and then its status. Schedule I on the 2023 return names Django Events Foundation North America as a grant recipient, gives its EIN as 073151229 and classifies it as a 501(c)(3). The registered EIN for that organisation is 47-3151229, which differs in the first digit, and the public record confirms it is indeed a 501(c)(3). The 2024 return names the same organisation again, for $6,000, but this time classifies it as an "other organization" rather than a 501(c)(3), and reports no EIN and no IRC section at all. The 2023 classification was right and the 2024 one is not.

A cross-reference to an explanation that is not there. The 2024 return ticks the Schedule O box for Part IX, which is the pointer that says "the breakdown is in the supplemental schedule". Schedule O on the 2024 return has four entries and none of them is about Part IX. The 2023 return ticks the same box and does provide the entry, spelling out the Fellow line as programme service expenses of $221,179. The disclosure matters more in 2024 than it did in 2023, because the line it would explain is 78.56% of the entire cost base, and it is the one year it is missing.

None of these three changes a headline figure. They are the kind of thing that turns up when a volunteer board, a single outside preparer and a cash-basis return meet a schedule as fiddly as Schedule F.

The donor base is concentrating

Schedule A tracks five years of gifts and runs the public support test. The DSF passes comfortably, but the trend underneath is real.

Those five sum to $1,318,180, exactly as reported. The four years that appear on both the 2023 and 2024 returns agree to the dollar on all four, which is the check that catches a misread column before it becomes a wrong table.

The public support percentage was 100.00% for 2022, 89.18% for 2023 and 84.20% for 2024. The threshold is 33.33%, so there is no risk here at all. The movement comes from Schedule A's substantial contributor line, which measures the amount by which large donors exceed 2% of total support, a threshold of $26,627 for 2024. That excess was $129,082 in 2023 and $197,161 in 2024, a rise of 52.74%. More of the DSF's money is coming from fewer, larger donors each year. At this scale that is a resilience question rather than a compliance one.

Who is answering this question today

Worth grading the incumbents separately, because they are not equally wrong.

The public summary record is correct and current. Its page renders the 2024 figures exactly: revenue $296,584, expenses $277,682, net income $18,902, net assets $274,497, contributions $289,456 at 97.6%, investment income $6,912 at 2.3%, inventory sales $216 at 0.1%. Every one matches the filed return. If you want the headline numbers, that page has them right, and the DSF's own annual report links to it.

Its API is a different matter, and this is the third organisation in this series where the same split appears. The machine-readable filings list stops at 2023, while the same record's tax_period field reads 2024-12-01 and latest_object_id names the 2024 filing that the list does not contain. The "filings without data" list is empty, so the 2024 return is absent from both. Anyone reading the page sees 2024. Anything iterating the filings list gets 2023 and no indication that a newer year exists.

Wikipedia is silent, not stale. The article contains no revenue figure and no monetary amount of any kind. That is a gap, not an error, and it is a fair thing for an encyclopaedia to leave alone.

The aggregators are the problem. Searching the DSF's revenue today, Google displays a RocketReach result stating that "The Django Software Foundation annual revenue was $5.5 million in 2026". The last filed figure is $296,584. That is 18.54 times the real number, and more than five times the $1,000,000 the DSF itself describes as an aspiration. The page returns a bot challenge to a plain request, so that figure is quoted from the snippet Google itself displays, checked on 13 September 2026.

Finally, the DSF's own foundation page is the top organic result for this question, and it contains no monetary figure at all and links to no financial document. Neither does the donate page, nor the site homepage. There is no index page for the reports: the obvious URL for one returns 404. The 2024 report is genuinely discoverable, through the June 2025 weblog post and the community forum, so this is not a transparency failure. It is a navigation one, and it is why an aggregator with a fabricated figure is competing for the answer at all.

Governance, briefly

Seven voting members, all seven independent, each reporting 1.00 hour per week, and $0 of reportable compensation for every one of them. No officer, director or key employee received anything. There is a conflict of interest policy and it is monitored annually. There is no whistleblower policy and no document retention policy, and the 2024 return reports that the organisation does not contemporaneously document the meetings of its governing body, where the 2023 return said it did. Volunteers fell from 15 to 7. The 2023 return puts the membership at about 210 people, whose main function is electing the board.

Both returns were signed by the treasurer, Jacob Kaplan-Moss, and prepared by the same CPA in Seattle, holding the same preparer identification number, at firms carrying two different names and two different EINs across the two years.

What this adds up to

The Django Software Foundation raised $296,584 in 2024 and spent $277,682, and if you stop there you have a number and no understanding. The structure is the story: 97.60% of income is donations, 78.56% of spending is one contracted programme, 84.30% of spending is booked abroad, there are zero employees, there is no fundraising cost, and there are 360.8 days of unrestricted reserve behind it.

The DSF's own ambition is spelled out in the same report. It met a $200,000 fundraising goal in 2024, and it describes what more would buy: at $500,000 an executive director and an additional Fellow, at $750,000 a doubled Fellowship and dedicated feature grants, at $1,000,000 larger event grants and a hardship fund. Against a filed base of $296,584, the first of those is a 1.69 times increase in annual fundraising, to hire the first person who would work on getting there.

For a comparison at similar scale but a very different structure, the Software in the Public Interest numbers are worth reading next: a fiscal sponsor whose balance sheet is mostly other people's money, where the restricted and unrestricted split is the whole analysis rather than a footnote.