🤠 Happy Wednesday. We must start with a huge congratulations to 🇺🇸 Reid Wiseman, 🇺🇸 Victor Glover, 🇺🇸 Christina Koch, and 🇨🇦 Jeremy Hansen. The Artemis II crew — the first humans to fly around the Moon since Apollo 17 in December 1972. After more than half a century, we are back.
IN THIS WEEK’S EDITION:
🤖 Omakase robots
☢ “Nuclear renaissance”
🧑🚀 Artemis II (we’re back, plus a DG clapback)
🪖 The DAWG, Hail Mary, & hypersonics
🌍 In memoriam: CIA’s World Factbook
Forwarded this email? Subscribe to Per Aspera here.
TWO WEEKS AGO… A four-foot-tall robot walked the night ward of a Japanese hospital. As part of Japan’s first humanoid clinical deployment, the robot guided patients to blood collection during the day, ferried lab specimens between floors, and patrolled corridors after outpatient hours. The three-day trial at University of Tsukuba Hospital ran on a Unitree G1 — which starts at $16,000 in the base configuration — and an OS called Omakase — Japanese for “I’ll leave it to you,” the phrase you use when you trust the chef.
It didn’t insert an IV or diagnose a patient — maybe one day — but walked a hallway, carried a tray, and stayed out of the way…
The hospital director watched the final demo personally, and staff praised the smoothness of its movement: the robot didn’t fall, didn’t collide with anyone, and nurses didn’t have to adjust a single thing.
But for all the fanfare, humanoids have largely yet to show they can go beyond a narrow range of tasks in real-world deployments. Where deployed, they’re typically doing one thing on a loop — a single station, a single pick, a single corridor. The state of the art is still a generalist platform deployed as a specialist.
A brief history of Specialist vs. Generalist
We’ve all at some point asked ourselves: What do I want to be when I grow up? Do I want to go deep or wide (domain expert, or T-shaped)? A similar dynamic has run through two of the major technology transitions of the past 40 years.
ASICs dominated compute for decades before the arrival of the GPU, originally a specialized chip that generalized and ate all of the AI epoch’s most important workloads.
Same story with AI… It used to be that you’d build separate, narrow models for every specialized task. Then, the transformer arrived and a single architecture absorbed the jobs of dozens.
Now, we’re seeing this play out as an intellectual holy war in the field of robotics: Team Specialized Bots v. Team Humanoid.
History doesn’t repeat, but it rhymes…
There’s a cycle we can observe from past transitions: the specialist arrives first, the generalist conquers the market, and the specialist reclaims the economics for some meaningful share of valuable work (see: Google’s TPU). Today, we ask: how might this story play out for robotics?
The humanoid is the ultimate generalist bet. But 60 years of mechanization history tell us that the physical world punishes generality in a way software never did. A crossover point appears to be getting closer — the body is already relatively “cheap” and getting cheaper, and the brain is getting smarter by the day. So is the humanoid destined to win out over specialized bots? Or is that a false choice?
A “NUCLEAR RENAISSANCE”…. In May 2020, a German utility detonated explosive charges and brought down both cooling towers at the Philippsburg nuclear power plant — a fully viable 1,402 MW station that had been switched off by law just five months prior, on New Year’s Eve 2019, in a Green-governed state. Five years later, for an encore: 600 kg of TNT, threaded through 1,800 boreholes, brought down the two towers of Gundremmingen in front of 30,000 spectators in Bavaria. Billed as a “visible symbol of Germany’s nuclear exit,” it was purely theatrical, as the reactors had been dead for years. Made for great TV, though.
Credit: Reuters
F-A, meet F-O … Over the preceding decade, Germany denuclearized and, in so doing, unilaterally disarmed itself of sovereign electrons. The country shut down 17 reactors that once provided a third of its power, starving its industrial base, increasing carbon intensity vs. the keep-them-running counterfactual, backfilling the gap with coal and imported gas, and then FAFO’ing when the gas was weaponized by Moscow, or more recently, held up in the Strait of Hormuz.
And now? A couple weeks ago, Germany’s energy minister called the nuclear phase-out “a huge mistake.” The same could be said of other rich, industrialized societies whose rulers talked themselves into denuclearizing in recent decades — and are now scrambling to memory-hole this choice and reverse course ASAP.
Something is happening everywhere all at once…
The Energy Department has been bull-posting about the “nuclear renaissance” on main for months now…. And last week, if you keyword-searched the phrase, you would have gotten hits from the Texas Governor, DOE, and a Wall Street bank.
A new global nuclear consensus has taken hold, as was made evidently clear over the span of ~100 hours last week:
NEW ENGLAND: On Tuesday, the region’s six governors signed a joint commitment to keep their existing reactors online and advance new sites.
TEXAS: On Wednesday, Lone Star State opened applications for a $350M fund to pull reactor manufacturing and fuel-cycle capacity into the state.
CALIFORNIA: On Thursday, Diablo Canyon, once slated for closure, won NRC license renewal through 2045.
(Ryan here…folks, as someone who's spent 90+% of his life in Texas or New England, trust me — you don't see these two places agree on much.)
Also within the last week:
JAPAN: Under Sanae Takaichi, Morgan Stanley writes that Japan's Nuclear Power Renaissance Begins: the new PM has made nuclear the centerpiece of her energy agenda, targeting a 20%+ energy share by the 2040s and putting advanced reactor new-builds on the table for the first time since Fukushima.
SOUTH KOREA: Kori-2, a 650 MW reactor that had been offline for three years, was restarted over the weekend.
INDIA: PM Narendra Modi announced this Tuesday that the first Indian-built breeder reactor has achieved criticality. (Breeder = a reactor design that produces more fissile material than it consumes, effectively manufacturing its own fuel.)
What happened?
Diablo Canyon, California’s largest clean power source, will remain online for two more decades.
The 180° turns in Berlin, Brussels, Sacramento, Washington, and elsewhere all amount to an admission that the denuclearization agenda was driven by something other than decarbonization, that clean baseload power shouldn’t be taken for granted, and that these rich societies will now pay the price to rebuild the thing they talked themselves into blowing up (in some cases, literally).
Even as you blow up a good thing you had going, others will keep building while you’re crashing out. China approved 10+ new reactor starts per year for four straight years and now accounts for 62% of all new construction globally — building at ~$2.50/watt in five years, versus ~$15/watt and a decade at Vogtle, America's only new-build this century.
THE GOOD NEWS: America still operates the world’s largest civilian reactor fleet, has the deepest capital pools, and there are price-insensitive buyers (hyperscalers) lining up to take whatever they can get, as the AI Supercycle, electrification, and reindustrialization pull grid growth out of a decade-long flatline. Last but not least, the political will to build has clearly arrived. Now for the hard part of rediscovering the lost art (and tribal knowledge) of building big reactors on schedule and on budget.
The biggest win from the successful Artemis II launch last Wednesday — in addition to closing a 53-year gap in crewed lunar missions — may be that it’s been pulling human spaceflight back into the center of American culture. The launch coverage livestream reached over 10M peak concurrent viewers across platforms, making it NASA’s most‑watched online launch to date, surpassing the previous record set by the 2020 NASA–SpaceX Crew Dragon Demo‑2 mission.
How we all watched the Artemis launch.
Even so, we as a community still have work to do. That Artemis II paused 10M+ of us to share in the moment together is an extraordinary achievement in the streaming era. But the bar history set for us is even higher: Apollo 11’s first moonwalk drew ~600M live TV viewers worldwide — roughly 1/5 of humanity at the time — making it the largest space broadcast in history. Yes, that was a very different media environment: a handful of dominant TV networks, almost no alternative channels, and an audience that was effectively funneled into a shared global moment. But with far more tools, platforms, and creative freedom than the Apollo generation could have imagined, why would we let ourselves aim for anything less?
Now would also be a good time to remember that NASA’s institutional memory is an undervalued American asset.
Some on the internet, less impressed with Artemis II, may say:
It's "extremely odd" that America landed on the moon in 1969, yet here we are celebrating a lunar flyby while China has a base on the dark side.
"NASA has been surpassed by SpaceX, Blue Origin, and the private sector."
But to them, our very own Dan Goldin responds…
So, whether you’re a skeptic or facing a skeptic, here are the facts (by Dan):
“China has only a modest robotic presence on the dark side of the moon — not a human base. They have never orbited the moon with astronauts.”
“SpaceX and Blue Origin have excellent launch vehicles for payloads to earth orbit — but neither has sent a human crew of four around the moon, let alone landed and returned.”
“Artemis II is a cutting-edge warm-up mission — designed to land four astronauts on the moon and bring them back safely. Apollo only had room for two.”
“Yes, there are warts and blemishes in Artemis — a dangerous return from lunar orbit hasn't happened yet, and we haven't landed on the moon. But give NASA a break for taking on the hard tasks!”
And one more from Babak Raeisinia (Cofounder, Machina Labs): “Folks forget that SpaceX and Blue Origin built and continue to build on the sound scientific and technological foundation put in place by NASA... dismissing NASA is failing to comprehend history and what it takes to achieve these feats.”
Human spaceflight is dangerous and costly. But it gives all of us the thrill of exploration and adventure — vicariously living the danger with the astronauts in real time, and taking pride in a nation willing to do hard things. That’s thanks to the people at NASA!
001 / UnderDAWG NO LONGER… The Defense Autonomous Warfare Group — yes, DAWG — was quietly stood up inside the Pentagon last year with a $225M budget. In the Trump administration’s $1.5T FY2027 defense spending proposal, released last week, the Pentagon requests $54.6B for the DAWG — a 24,167% YoY increase — that nearly matches the entire Marine Corps budget. The DAWG’s remit includes bigger autonomous weapons like one-way attack drones (Shahed-class & LUCAS) and uncrewed attack boats (Ukraine’s Black Sea playbook, Hellscape), while Drone Dominance retains the smaller FPV portfolio. A “group” typically doesn’t get $55B, but a command might. Russia formalized its own unmanned systems force this year with 210,000 personnel. So, maybe the DAWG won’t stay a group for long…
002 / BTW… Dan wants you to go see Project Hail Mary on the big screen: “Don’t wait for it to come out on Netflix. And just overlook the wobbly science. It’s about the human experience — about friendship, connection, and choices made under extreme pressure.”
003 / HYPERSONIC FLURRY… This past week was a hypersonic flurry: the U.S. Army and Navy tested a Mach 5+ weapon, Turkey publicly showcased and tested its Tayfun ballistic missile system, India highlighted a successful long‑range hypersonic missile test underscoring its drive toward a future hypersonic arsenal, and the chief designer behind Russia’s Zircon system, Alexander Leonov, died. (And oh yeah, Hermeus announced a $350M Series C led by Khosla Ventures). Of course, this is a domain where we’ve got to separate showmanship from execution, but nonetheless there’s a lot happening here. To get grounded on hypersonics, read our hypersonics antimemo in collaboration with the “Godfather” of the technology Mark Lewis.
So long, World Factbook… Born classified in 1962 as a cheat sheet for spooks, the CIA’s World Factbook spent six decades evolving into one of the internet’s most reliable free reference tools. Then in February, it vanished as quietly as it arrived. Langley sunset the resource without any warning, archive, or replacement — and in its place, offered a new landing page reminding visitors to “stay curious.” Pour one out, not for the intelligence analyst who needed Tajikistan’s GDP, but for all of the 7th graders who needed Tajikistan’s GDP the night before their essay and bibliography were due. In the age of LLMs, they’ll manage.…
|