Most weeks I send you the 1% of AI and tech stories that move business. This week I'm doing something different.
A reader sent me a stack of science and health breakthroughs and asked which were real. I checked all of them, and the honest answer is: most are real, several are oversold, and a couple are quietly more important than the viral version.
So this is a special issue. Normal programming resumes next week. The thread running underneath, the one that ties it back to our usual beat, is at the bottom.
Medicine
A pill for pancreatic cancer got a standing ovation from a room of doctors who almost never clap. At the big oncology meeting in Chicago on May 31, researchers presented daraxonrasib, the first drug of its kind to work against pancreatic cancer, and the data doubled median survival from 6.7 to 13.2 months.
It blocks a mutated protein that drives more than 90% of these tumors, a target that resisted treatment for decades. This is the freshest item on the list and the one that matters most: pancreatic cancer kills most patients within a year, so adding six months is not a footnote.
It is not approved yet. Full FDA clearance could land later in 2026, and it is already available through an expanded-access program for some patients now.
Scientists grew insulin-producing cells that reversed diabetes, in mice. A team at Karolinska reversed the disease in lab animals using insulin cells grown from stem cells, and a separate group built the first lab-grown islet "mini pancreas" with its own blood vessels, which is the part that usually kills these implants.

The viral framing calls it a cure for diabetes. It isn't, yet. It is a real step toward replacing daily injections with a one-time cell transplant, and the blood-vessel problem was the wall everyone kept hitting. Watch for the first human trials, which is where this either becomes medicine or stays a paper.
France's "forever" artificial heart is real, and the company that made it just died. The Aeson total artificial heart from French firm Carmat works, runs without a donor, and has been implanted in patients for years as a bridge to transplant. But it is rated for roughly five years, not forever, and Carmat filed for insolvency and was ordered into liquidation in January.
That is the part the feel-good posts leave out, and it is the real lesson: the engineering can succeed while the business fails. Breakthrough hardware needs a company that can survive long enough to ship it.
Defense Spending Is Surging. Here's Where It's Going.
Global defense budgets are expanding, but the allocation has changed. A growing share of spending is going toward AI-enabled systems, satellite networks, and advanced aerospace, not the platforms that dominated the last generation of procurement. We identified five companies at the center of this reallocation in a single research brief. Inside, you'll find the investment case for each, the contracts driving revenue, and the risks worth understanding before you commit capital. If you want exposure to defense sector growth beyond the traditional mega-caps, this report is a practical starting point. Free, concise, and built for investors who want to move ahead of the crowd.
Vision
Surgeons restored sight to a legally blind patient with a cornea printed from cultured cells, a world first. In late October, the Rambam Eye Institute in Israel transplanted a 3D-bio-printed cornea grown entirely from human corneal cells, the first time anyone has done it in a person.
The number that matters: a single donated cornea was cultured to print around 300 implants. There is currently one donor cornea available for every 70 people who need one, so turning one into hundreds is the whole game. Early days, one patient, but this is the kind of supply math that changes who gets to see.

Eye drops that replace reading glasses are real, and already on sale. The viral claim says they arrive by end of 2026. The truth is they have been here since 2021 (Vuity), with Qlosi and VIZZ both approved in 2025 and a new combination drop awaiting an FDA decision.
The catch worth knowing: they only fix age-related near vision, the kind that sends people over 40 reaching for readers. They do nothing for nearsightedness, so they don't replace glasses for most people who wear them. If you have started holding your phone at arm's length, ask an ophthalmologist. If you are nearsighted, keep your glasses.
Infrastructure and everyday life
Japan pushed a single fiber to 1.02 petabits per second. A team led by Japan's NICT set the record over 1,808 km using a 19-core fiber that fits inside standard cable. That is about four million times the typical home broadband speed, hence the "download all of Netflix in a second" line going around.
Keep one thing straight: this is a lab record, not something coming to your router. It matters because the fiber matches existing infrastructure dimensions, so the path from lab to backbone is shorter than usual. This is the plumbing the AI buildout will run on.

South Korea is testing walking lanes for slow and fast pedestrians. Stations and busy streets are trialing color-coded lanes, a fast stripe and a slow stripe for the elderly, tourists with luggage, and parents with strollers.
The goal is less collision, less irritation, lower street-level stress. It is the softest item here and the most human. Sometimes the upgrade isn't a chip, it's a painted line that lets a grandmother and a commuter share a sidewalk without resenting each other.
And from our usual world
Anthropic filed to go public, and admitted its AI now writes most of its own code. The maker of Claude filed a confidential draft IPO on June 1 after a raise that valued it near $965 billion, and in the same week disclosed that more than 80% of the code merged into its own codebase is now written by Claude. Read that next to everything above.
The pancreatic-cancer target was found with computational tools. The protein-folding work behind half of today's drug pipelines is AI. T
he reason this was a medicine week and not a software week is partly that the software got good enough to point at the body.
When a lab tells you its product builds the next version of itself, the breakthroughs in the operating room are the same story told one layer down.
Next edition soon, Çelik
Viva Technology - June 17-20, 2026, Paris, France
The Next Web - June 19 - 20, 2026, Amsterdam, Netherlands
We Are Developers - 8 - 10 July, 2026, Berlin, Germany
Bits & Pretzels - September 18-20, 2026, Munich, Germany
How to Web - October 7-8, 2026, Bucharest, Romania
Web Summit - November 9 - 12, 2026, Lisbon, Portugal
Slush - November 20-21, 2026, Helsinki, Finland


