Founders · Value created vs. wealth kept
Rank founders by the value they created for others, not the wealth they kept. Each ledger sums what a company generated for workers, taxpayers, shareholders, and customers, then compares it with what the founder kept. Financial data comes from SEC filings. Every estimate is labeled.
"Somebody needs to make a list where they rank people by how much wealth they've created for other people. … That should put me pretty high on some kind of list, and that's a better list."
Jeff Bezos, New York Times DealBook Summit, December 4, 2024. This site is that list.
Eight founders scored across seven companies. Every case uses one method, so you can compare them. Customer surplus is the exception: each industry gets the best published research available for it, and for the two ad-funded companies that research disagrees about the sign. Amazon had the deepest check, with 25 claims tested by independent reviewers. The rest had one check per input.
Critics say someone else would have built it. Suppose they are right, and give Bezos only part of the credit. These scenarios use the he kept against the mid-case total.
Each bar segment is one stream of value that flowed to someone other than Bezos. The thin bar below it is what he kept, drawn at the same scale. We checked every input against primary sources in September 2026. We tested 25 claims. 24 held. One failed.
Cumulative value created, 1996 to 2025 (mid case)
Billions of dollars. Hover a segment for detail.
Both series come straight from Amazon's 10-K filings. Headcount grew from 256 people to 1,576,000. Revenue grew from $16 million to $717 billion.
Employees at year end
Source: 10-K filings via SEC EDGAR
Annual revenue, $B
Source: 10-K filings via SEC EDGAR
"Actual" means the number comes from an SEC filing. "Estimate" means we computed it from disclosed data plus a stated assumption. "Literature" means the range comes from published economics research.
In September 2026 every Amazon input went through review: 22 sources fetched, 25 claims tested by three independent reviewers each, 24 confirmed and 1 refuted. A second review on September 16 removed a double count. These are the revisions.
Every case uses the Amazon method: pay anchored to SEC pay-ratio disclosures, cash taxes from SEC XBRL, a dated stake, a customer surplus from published research, and a cost side. Open one to see its streams, their sources, and what changed in review.
Not by itself. It measures how much of the value a founder kept, not whether they deserved it. Nordhaus (2004) found innovators across the whole US economy keep about 2.2% of the value they create, range 1.3% to 3.3%. That is the only benchmark here. A founder above it kept more than the historical average, and may still have passed on the great majority of what they created.
It used to. Kept now counts past sales and gifts as captured, so a founder who sold or gave shares away is still charged for them. That matters most for Gates, who has given away more Microsoft stock than he still holds: counting only what he holds today would put him at 0.6% instead of 1.9%. One weakness remains. Shareholder value dominates every total, so the capture share still tracks the founder's ownership stake closely.
Because the question is how much of the value a founder captured, not what they did next. Giving is a use of captured wealth. The Gates Foundation has done a great deal of good with $64B, and none of that changes how much of Microsoft's value ended up with Gates rather than with someone else. The same logic excludes taxes the founder paid on a sale: those already sit in the taxes stream.
The sale proceeds that funded it are counted, at the price Bezos got for the shares. What Blue Origin has done with that money since is Blue Origin's value creation, not Amazon's. Counting the later valuation would credit Amazon for a different company's work. The same rule keeps SpaceX and xAI out of Musk's figure and Cascade Investment's gains out of Gates's.
Then the founder deserves less credit. Use the credit control above to discount it. At 25% credit, every capture share roughly quadruples. The ledger takes no position on the right number.
For Nvidia, largely yes: 94% of its total is the share price, so its ledger is mostly a bull market. Tesla is 79%, Microsoft 63%, Meta 57%, Amazon 56%, and Alphabet 54%. That is why the site reports a second ratio with the share price removed. On that measure Microsoft returned $19.40 to workers, taxpayers, and customers for every dollar Gates kept, Amazon $5.80, Alphabet $5.70, Meta $4.50, Nvidia $1.70, and Tesla about $1. It is a different ranking and it does not move with the market.
Customer surplus, then indirect wages, then Gates's sale proceeds. No published study measures any of these companies on its own, so each case scales a result from the nearest research. Indirect wages rest on a multiplier no study confirms or refutes. Gates's $34B of past sales is a reconstruction from share counts and split history, with a $25B to $45B band, because his largest disposals predate electronic filing. The first two have controls above.
Revenue, headcount, stock compensation, cash taxes, and share counts come from SEC filings, mostly through the XBRL company-facts API. Pay comes from proxy pay-ratio disclosures. Stakes come from proxies and Schedule 13 filings. Net worth comes from Forbes real-time on September 15, 2026. Customer surplus comes from published economics research, named in each row. Every row says which of those it is.
Compare · Side by side
Same method, same scale, every case. Choose who to compare and which number to compare them on.
Total value created for others (mid case)
Billions of dollars
The share price is most of every ledger. How much is left once you take it out is the fairest comparison between these six, because it does not move with the market.