Inside Vanguard's alarming state of retirement in 2026

For 25 years, Vanguard has been keeping score on the great American retirement experiment. Every year, it publishes the most comprehensive snapshot in the industry of how workers save, how much they accumulate, and whether any of it will be enough. The 2026 edition is out. Spoiler: It depends on which number you’re looking at.

The number Vanguard leads with is $167,970—the average 401(k) balance across nearly 5 million accounts the firm administered at year-end 2025. That’s a record high, up 13% from the year before—carried there on the back of a stock market that had a very good year.

The number sitting next to it is $44,115. That is the median—the balance belonging to the worker planted exactly in the middle of the distribution, with half of Americans above and half below. Run it through a standard 4% annual withdrawal rate and you get $1,765 a year, or $147 a month. That’s not enough to cover rent, a month of blood pressure medication or, in any honest accounting, fund a retirement.

The distance between $167,970 and $44,115 is the architecture of American inequality, expressed in retirement savings data—a small cohort of high-balance savers dragging the average skyward while the rest of the country lives in the median.

And the median is getting squeezed from both ends. Even as balances climbed to those record highs, Americans set another record in 2025: 6% of Vanguard participants made a hardship withdrawal last year, up from 5% in 2024, triple the pre-pandemic rate, and the sixth straight annual increase. The median withdrawal was just $1,900, showing withdrawals were the only option they had.

Against what Americans believe they actually need, the math becomes almost darkly comic. Northwestern Mutual found this year Americans think they need $1.46 million to retire comfortably. Fidelity says you should have 10x your salary saved by 67. For a $60,000 earner, that’s $600,000. That means the median retirement account is behind by a factor of 13.

So whose fault is it? Here is where Vanguard’s own 25 years of data become uncomfortable for the system it serves.

The answer is not the worker. Plan design—automatic enrollment, automatic escalation, professionally managed default funds—is the variable that actually determines retirement outcomes. Workers who get nudged into saving, save. Workers who get defaulted into staying invested, stay invested. According to a June 2026 Investment Company Institute survey, nearly half of all Americans with a 401(k) say they probably wouldn’t be saving for retirement at all without one. The plan isn’t supplementing individual initiative. For most people, it is the only reason retirement savings exist.

This means the millions of Americans whose employers offer no plan—or a gutted plan with none of the automatic features—aren’t making a bad choice; they’re being handed a bad deal. Teresa Ghilarducci, the New School economist who has spent her career cataloguing the failures of the voluntary retirement system, put it plainly to Fortune earlier this year: Low-income workers excluded from these plans for decades doubt the payoff of a 401(k).

Many low-income earners “really want me to sit down and explain how it worked for them, because they’ve just been excluded from a system like this for their whole careers,” she said. “They want to know what the catch is.” 

Vanguard recommends saving 12% to 15% of annual pay, including the employer’s share, to stay on track. For a $60,000 household already managing rent, childcare, student loans, and health care, most Americans are nowhere near it, and the system is not designed to get them there.

After 25 years of data, a 33-to-1 gap against what people believe they need, six straight years of rising hardship withdrawals, and a retirement architecture that still leaves tens of millions of workers outside the door, Vanguard’s most unsettling finding may be the simplest one: we have known about this problem for a very long time. We just haven’t fixed it.

Vanguard’s How America Saves 2026 is based on data from nearly 5 million defined contribution plan participants across Vanguard-administered plans as of year-end 2025.

For this story, Fortune journalists used generative AI as a research tool. An editor verified the accuracy of the information before publishing.

Nick Lichtenberg is business editor and was formerly Fortune's executive editor of global news.