Did Ozempic slow aging, or did the lost pounds? None of this is peer reviewed yet. Novo Nordisk's semaglutide data is a preprint, and Eli Lilly's tirzepatide data was a conference talk at ARDD in Boston last week. Novo pooled blood from 10,052 people in five placebo-controlled trials and ran it through proteomic clocks, blood-protein scores trained to estimate each organ's risk of dying. Heart age came out about 2 to 4 years younger on the drug. These were high-risk patients from heart and kidney trials, followed for up to three years. Lilly's version is thinner. Per a write-up of Kevin Duffin's talk, the tirzepatide data comes from SURMOUNT-5, and the clocks were checked against the calendar. Over those 1.38 years, all 15 epigenetic clocks ticked forward by less than the calendar did. SURMOUNT-5 ran 72 weeks against Wegovy with no placebo arm, so no untreated group existed to compare against. MIT Technology Review described both results as drug against placebo, and none of the coverage I read notes that Lilly's trial had no placebo to beat. Neither data set measured how long anyone lived. SELECT, the largest of Novo's five trials, cut major heart events by 20%, and that is a real outcome. The clock is a forecast of outcomes, and the preprint's authors write that their result does not establish rejuvenation. Weight loss can move aging clocks too. In 22 bariatric surgery patients, one DNA-based clock showed 4.29 fewer years of excess aging within a year of the operation. Novo's preprint says weight loss does not fully explain its effect, a claim no reviewer has checked. Tirzepatide brought Lilly about $36.5 billion in 2025. Per the same write-up, one of the two labs that replicated Lilly's clock readings was TruDiagnostic, which sells epigenetic age tests to consumers. Vadim Gladyshev, who helped Novo with the measurements, told MIT Technology Review that for healthy people, no one knows whether it is an anti-aging drug. The trial built to answer that, VITAL-H at UT Health San Antonio, will test semaglutide and two other approved drugs against placebo in 726 adults in their 60s. ARPA-H is paying up to $38 million, and enrollment is planned for 2027. What to check next is whether the Novo preprint survives peer review with its weight-loss adjustment intact. Until then, the drug half of those 10,052 patients looks younger than the placebo half on a clock built to predict death, and that is as far as the data goes.

Observational data: wealth falls six years before dementia. The study is an NBER working paper, not yet peer-reviewed, so treat it as early evidence. It tracked 2,312 Americans classified as developing dementia and 8,431 who didn't, using survey data from 2000 to 2016, and the two groups' wealth starts to split about six years before onset. At onset the gap is about $124,000, and about 63 percent of it is financial wealth, mostly money held in investment and bank accounts. Eight years out the groups differed by about $30,000, a gap too small to be statistically meaningful, and it barely moved until year six. Earnings barely move: at onset the gap is $947. Before onset, out-of-pocket medical bills run about $1,000 to $1,600 a year higher, too little to explain it. Retirement accounts like IRAs, which are hard to touch, show no relative change, and none of the press coverage I read mentions that result. Onset here comes from a statistical model built on memory test scores. No doctor made that call, and the NBER's own summary still puts "before dementia diagnosis" in its title. It fits earlier work. A JAMA Internal Medicine study of 81,364 Medicare beneficiaries found missed bill payments up to six years before diagnosis, though the absolute gap at that point was 7.7 percent against 7.3 percent. Brokerages already have rules for this. Since 2018 FINRA lets them put an initial hold of up to 15 business days on a payout from the account of a customer 65 or older if they suspect someone is exploiting that customer. Account-watching subscriptions for older people start around $7.49 a month. Healthspan is usually told as a medical story. Mayo Clinic researchers put the US gap between lifespan and healthy lifespan at 12.4 years, the widest of 183 countries, and this paper suggests part of the cost lands in household money years before onset. The six-year timing looks solid; the cause is inferred by ruling out alternatives, as the authors write themselves. The money that held up was the money that was hardest to get at.

Growth signaling cut, female mice lived 8 percent longer. Mouse study, out this September in Aging Cell from John Kopchick's lab at Ohio University. The receptor for growth hormone was switched off at 12 months, roughly middle age, in 30 to 35 animals per sex per group. In females, median lifespan rose 8 percent, from 929 to 1,003 days. The p value was 0.049, just under the usual 0.05 cutoff for calling a result significant. Males gained 7 percent, which did not reach significance on that measure. The paper's title says it extends healthy lifespan. In females, the healthy part did not show up: no gains in grip strength, balance, frailty or bone. Males improved on insulin sensitivity, grip and balance, and both sexes carried more body fat. Anti-aging clinics have sold growth hormone injections for years, pushing that signal up. These mice lived longer with it turned down. In the US, distributing growth hormone for anti-aging is a federal felony. The classic version, mice born without the receptor, has been one of the most replicated lifespan results in mammals since the late 1990s. Switching the receptor off in midlife, the only timing a human drug could realistically copy, gives a much smaller effect. Humans with the same receptor defect exist. In Ecuador, about 100 people born with it (Laron syndrome) were followed for 22 years, and none developed diabetes, against 5 percent of their relatives. Cancer appeared once, and it was not fatal, against cancer in 17 percent of relatives. They did not live longer, mostly because of accidents, alcohol and other causes unrelated to age. This is observational data, not a trial. So the pathway with nearly three decades of mouse data, switched off in midlife, produces a modest, borderline result. Meanwhile the field's hottest idea, cell reprogramming, got New Limit $435 million in June at a $3.1 billion valuation, with no human data yet and a first trial planned for 2027. A drug that turns growth hormone signaling down in midlife without the extra fat would be a real business. That is a concept. Nobody has one built for aging yet. A modest effect in one cohort of mice, with a cost attached. An 8 percent gain in one sex at p equal to 0.049 is a reason to repeat the experiment, and not a reason to buy anything a clinic is selling.

Dead bacteria made old mice look younger on paper. Nature Aging published the study on September 24, 2026. It comes from Jing-Dong Han's lab at Peking University, and the organism is Neisseria flavescens, a bacterium that already lives in your mouth. No human picked this bacterium. The team built an AI aging clock they call AURORA, had it model what would happen if you added or removed each microbe, and this one came out on top as the predicted winner. Screening by model is normal practice and fine on its own. What matters is what the screen got followed with. In people, nothing. The human data is associative, with no intervention at all. The closest thing to a longitudinal check tracked 132 participants over a year, and it is still a correlation. And in the study's own supplementary figures, when they ranked people directly by how old they looked biologically, N. flavescens did not come out as a top correlate. The headline exists because the model put it there. In worms, live bacteria did extend lifespan and healthspan. Movement was scored on 30 worms per group. In mice it gets thin. Heat-killed bacteria, ten weeks, 20-month-old animals, six per group and five in some panels. Blood chemistry and liver gene activity shifted toward the patterns of younger mice, and the gut bacteria shifted too. Survival was never measured, and there is no mouse lifespan curve anywhere in the study. Food intake was recorded per cage with no statistics run on it. The supplementary notes say the P values weren't corrected for testing many things at once, which is how findings that aren't there get published. The authors' own figure title calls the reversal partial. Worms got live bacteria and mice got dead ones, and the sections I can read don't say why the switch was made. That's the first thing I'd want answered. No company is attached to this, and no investor. The funding is Chinese public money. The one commercial fact in the study is a pending patent, number 202611084872X, filed by the senior author and a colleague and not yet publicly viewable. It was filed before a single human had been given anything. The selling started years ago regardless. Viome runs a $399 test that reads your oral microbiome and hands back a biological age score, plus lozenges at $79 a month on subscription. Pendulum sells Akkermansia muciniphila for $54, and Akkermansia turns up inside this very study as a loose correlation rather than a result. Bristle sells a $169 oral microbiome test and makes no aging claim at all, which deserves credit. Keep the two markets apart. The longevity economy means everything sold to older people, and it is enormous. Biotech that actually treats aging is a much smaller thing that mostly hasn't worked yet. The products above sit in the first market while borrowing the language of the second. There's a history here too. In 2018 this same bacterium was pulled from gastritis patients and written up in Current Microbiology as a possible pathogen, with the authors asking someone to study whether it causes disease. Nobody appears to have run that study, and nobody has established what a live strain does in a person. This is worth watching and not worth buying. An AI model nominated the bacterium, worms lived longer on it, and six old mice per group had younger-looking blood work after ten weeks of dead cells. That's a hypothesis with a patent attached, and the readout that matters is whether anyone runs it in living humans. Zero survival curves in the study. One $399 test already on the shelf.