68 Quadrillion ...

Society for the Protection of Underground Networks (SPUN)

Pandora has nothing on earth in terms of terrestrial connectivity as it seems, fungi is the connective tissue for all earth, capable of reaching star systems as described by a recent study published in Science .

Scientists have mapped the Earth’s entire underground fungal network, showing that it’s so extensive that if it were stretched into a straight line, it would reach other star systems — and span a sizable chunk of the Milky Way galaxy, for that matter.

The groundbreaking work, published in a study in the journal Science, focused on microorganisms known as arbuscular mycorrhizal fungi. Forming the hidden backbone of our planet’s soil, they circulate water and nutrients and regulate the climate by locking away vast stores of carbon. 

Altogether, the global fungal network weighs around 300 megatons, the study found, which is four to six times more than the biomass of all human beings. Around 40 percent of that fungal mass resides in high-altitude or flooded grasslands, like the Everglades in Florida. 

The authors hope that their work will highlight the indispensable but overlooked role that these fungal networks play in the Earth’s ecosystems, with around 70 percent of all ground-based plant life depending on the fungi.  

Fluorescently labelled carbon droplets, yellow, flowing inside hyphal network of arbuscular mycorrhizal fungi in real-time. Laid end to end, the filaments of these fungi would extend for 68 quadrillion miles. Credit...Achille Joliot/AMOLF Institute, Shimizu Lab

A hidden circulatory system pulses just beneath the planet’s surface. There, embedded in the soil, are dense networks of microorganisms known as arbuscular mycorrhizal fungi. These fungi attach themselves to the roots of plants, sending long, thin filaments out through the soil. These ferry water and nutrients to plants and whisk away carbon, helping to keep vast quantities of it out of the atmosphere.

Laid end to end, these filaments would extend for 68 quadrillion miles — roughly 730 million times the distance between Earth and the sun, scientists reported in a new paper. Collectively, the filaments contain approximately 300 megatons of carbon, or four to six times as much as the carbon contained by all the human beings on the planet, according to the study, which was published in Science on Thursday.

An international team of scientists conducted the research, using a combination of advanced techniques, including machine learning and a high-resolution imaging robot, to measure, predict and map the size of these fungal networks in ecosystems across the globe.

The study revealed particularly dense fungal networks beneath the world’s grasslands, adding to the evidence that these ecosystems, which tend to receive less conservation protection than forests, serve as significant carbon sinks for the planet.

The researchers also photographed lab-grown fungal samples with a high-tech imaging robot to make high-resolution images of the hyphae, allowing the scientists to determine their width.
Credit...Loreto Oyarte Gálvez/VU Amsterdam/AMOLF/SPUN

One thing not discussed is the fact Microtubules reside in all the fungi,
something to think about is it not?

Microtubules are part of the cytoskeleton, a structural network within the cell's cytoplasm. The roles of the microtubule cytoskeleton include mechanical support, organization of the cytoplasm, transport, motility and chromosome segregation. In developing neurons microtubules are known as neurotubules,[23] and they can modulate the dynamics of actin, another component of the cytoskeleton.[24] A microtubule is capable of growing and shrinking in order to generate force, and there are motor proteins such as kinesins and dynein that allow organelles and other cellular components (such as mRNA) to be carried along a microtubule, using specific adaptor proteins.[25][26]
This combination of roles makes microtubules important for organizing and moving intracellular constituents/cargo.

 It from Bit applies.

In 3 parts ...

3 part soliloquy

What is this? Being rather math challenged to the extreme, I asked Gemini to verify my explanation for the Einstein and Wheeler-DeWitt equations as I understand the first as C = light/Photon, a Boson, the force carrier particle equivalent to matter with the dividing line being the speed of light as that is the exact center point where the analog boson of light transitions into digital matter i.e. Photon transitions to a quanta to enter into the Electron. Think photoelectric effect to visualize why this is so. In effect, Einstein's equation depicts the analog to digital process from the energy side.

 
Note: Photons, gravitons and gluons travel at the speed of light.

As for Wheeler-DeWitt, it depicts the analog to digital transition as a balancing act, directly equivalent to Einstein in execution but different in application as it shows why the collapse of the wave function from analog to digital (decoherence) dissipates the energy of analog by transiting to digital. Think the E Major chord in
Day in a Life - Sgt. Peppers and Yang & Yin as the two equations are complementary.

Because the transition occurs at the speed of light, there is no time as the speed of light for gravitons, photons and gluons mean the three entities are timeless, hence the transition from analog to digital and vice versa is timeless.

I actually passed the test. :)

Yin and yang (/jɪn/ and /jæŋ/), yinyang,[1][2] or yin-yang[3][2] is a Chinese philosophical concept
describing opposite but interconnected forces. 

Quoting yours truly as backgrounder ...

The act of observation generates the collapse of the wave function as described by the Copenhagen interpretation of Quantum Mechanics, the physics of the very small, an Alice in Wonderland view of the universe writ large when seen under the guise of exquisitely designed experiments showing how this aspect of reality rings true. What's interesting is the fact the act of observation is often anthropomorphized and not viewed as a core element of how existence functions as everything's entangled and decohered outside of the realm of humanity, thus enabling us rubes to see a consistent view of existence save for the extraordinary experiments conducted by physicists obsessed with how reality works at deep level.

Wheeler's insightful notion of existence,  It from bit, being an information and transaction construct makes sense as the boson, (the photon), the force carrier of the electromagnetic, becomes digital by transitioning into discrete energy packets (quanta) when absorbed by leptons (the electron), a radical solution to the black body problem described in detail by Max Planck in 1900, thus introducing Quantum Mechanics to the world, the prime driver giving rise to the digital age in which we all live.  Prior to absorption,  photons, aka light, exist as analog waves, something readily seen by Thomas Young's brilliant 1801 double slit experiment.

Photons or matter (like electrons) produce an interference pattern when two slits are used.

The matter/wave rules

Position space probability density of an initially Gaussian state moving in one dimension z

at minimally uncertain, constant momentum in free space

When I conceived the first basic ideas of wave mechanics in 1923–1924, I was guided by the aim to perform a real physical synthesis, valid for all particles, of the coexistence of the wave and of the corpuscular aspects that Einstein had introduced for photons in his theory of light quanta in 1905.

— de Broglie[7]

Reality is the exchange of information, It from bit,  which means it is participatory. 

From John Wheeler (1983), Law Without Law. In: Quantum theory and measurement.

The particle zoo

The class of Bosons is one of the two fundamental classes of subatomic particle, the other being Fermions. All subatomic particles must be one or the other. A composite particle (hadron) may fall into either class depending on its composition.

3 part soliloquy

Why 3 parts?

Analog/the blur/digital

A repeat applies.

Because the transition, the center panel in 3 part soliloquy, occurs at the speed of light, there is no time as the speed of light for gravitons, photons and gluons mean the three entities are timeless, hence the transition from analog to digital and vice versa is timeless.

And so it goes - K. Vonnegut

Bessie/rev II

Bessie/rev II

12 years ago, Bessie was a blurb penned in 2014 describing the unsustainability of how beef, chicken and pork are produced in this country as the costs needed to support said enterprises boggle the mind.

In 2014 ...

Among the key findings in the report are:

• More than 1.7 billion animals are used in livestock production worldwide and occupy more than one-fourth of the Earth's land.
• Production of animal feed consumes about one-third of total arable land.
• Livestock production accounts for approximately 40 percent of the global agricultural gross domestic product.
• The livestock sector, including feed production and transport, is responsible for about 18 percent of all greenhouse gas emissions worldwide. 

In 2026

A single kilogram of conventionally farmed beef requires approximately 15,400 litres of water to produce, the equivalent of around 100 standard bathtubs, and generates approximately 99 kilograms of carbon dioxide equivalent in greenhouse gas emissions, the equivalent of driving an average car about 250 miles.

The figures come from two of the most carefully constructed peer-reviewed datasets in environmental science. The water footprint estimate is from Mesfin Mekonnen and Arjen Hoekstra of the University of Twente in the Netherlands, who published their global analysis in the peer-reviewed journal Ecosystems in 2012. The carbon footprint estimate is from Joseph Poore at the University of Oxford and Thomas Nemecek at Agroscope in Switzerland, who published their global meta-analysis in Science in 2018. The two studies together represent the strongest available scientific evidence on the environmental cost of beef production.

And we haven't even discussed Leghorn Foghorn or Porky but we will. In thinking about this, the connect to data centers, rockets and fossil fuel use, fits the pattern of extraction and continued abuse of planet earth. Gaia wants us gone and with global warming and unfettered use of AI, getting us gone is no longer in the realm of science fiction.

Channeling 2014 yet again.

Here are some more interesting facts on what goes on mankind's dinner table.

Livestock production — which includes meat, milk and eggs — contributes 40% of global agricultural gross domestic product, provides income for more than 1.3 billion people and uses one-third of the world’s fresh water. There may be no other single human activity that has a bigger impact on the planet than the raising of livestock. But as a new study out today in the journal Proceedings of the National Academy of Sciences (PNAS) shows, there is tremendous variation in how we raise livestock around the world — and major differences in what that means for the earth and for us. 

Sadly, the http://science.time.com/ site no longer exists but for BRT, it does. :)

The study of planetary habitability is partly based upon extrapolation from knowledge of the Earth's conditions, as the Earth is the only planet currently known to harbour life
(The Blue Marble, 1972 Apollo 17 photograph).

The ultimate heist

Futurism/Getty Images

The ultimate heist, subtle, addictive and frightening, summaries. Summaries of trending news stories like this one, a Futurism piece describing, with precision, how the heist is happening. "Wonderful" BRT doesn't do this. Colored type, pix with credits and links to the sources is an absolute because getting a Quisinart of the piece via AI in order for Google to basically control access to the news is disquieting to say the least. 

In the latest instance of Google discouraging users from clicking on links, the search giant has launched a new feature that uses AI to summarize trending news stories.

As first reported on by TechCrunch, the AI summaries are appearing for users in the US inside Google’s main mobile search apps on iOS and Android under the main Discover news feed. Tiny logos of publishers appear in the upper left-hand corner, haphazardly reminding users of where the information actually came from.

Underlining the janky nature of the tech, large text under the summary warns that it was “generated with AI, which can make mistakes,” seemingly in an attempt to preempt blowback of the type that Apple received after it rolled out a similar news-summarizing feature that turned out to hallucinate wild misinformation.

As follow up.

Image Credits: Google

Google Discover adds AI summaries, threatening publishers with further traffic declines.

As publishers fret about decreased traffic from Google, the search giant has begun rolling out AI summaries in Discover, the main news feed inside Google’s search app on iOS and Android. Now, instead of seeing a headline from a major publication, users will see multiple news publishers’ logos in the top-left corner, followed by an AI-generated summary that cites those sources.

The app warns that these summaries are generated with AI, “which can make mistakes.”

Image Credits: Google

The feature is not yet appearing for all news stories within the Google app. In tests, TechCrunch
was able to view the AI summaries firsthand across both iOS and Android apps in the U.S.

Reached for comment, a Google spokesperson confirmed this is not a test, but actually a U.S. launch.
The feature will appear on iOS and Android in the U.S., with a focus on trending lifestyle topics like sports
and entertainment. Google also noted the feature will make it easier for people to decide
what pages they want to visit.

95% chance people won't as people are lazy. One click instead of two always wins.

Just another way to get ever more information out of us rubes as it's all about the money.

A pissing contest :)

Calvin, forever :)

Calvin and Hobbs, integrity 110 forever. Calvin's masterpiece applies to the ongoing pissing contest of OpenAI and Anthropic as they go for IPOs to stanch their ongoing dumpster fires of data center expense, lack of any meaningful ROI and increasing awareness they're losing control of a tech they will never fully understand. With that in mind, let the contest begin. :) 

Corporations are reeling after finding that the cost to access powerful AI tools is soaring — despite showing no clear payoff. In one particularly unfortunate incident, according to Axios, the CFO of a company accidentally racked up half a billion dollars in Claude usage fees in a single month.

Shutterstock / Futurism

Put simply, the horrible economics of AI are finally starting to rear their ugly head. Astronomical capital expenditures by AI companies are starting to trickle down to users — and they’re not liking what they’re seeing.

Meanwhile, as the Wall Street Journal reports, executives at OpenAI are pondering whether to kick off a price war with the company’s biggest competitor, Anthropic. By dramatically lowering prices, the company’s reportedly hoping to steal users, while also anticipating similar price cuts by its competitor.

Put simply, pricing is turning into a major headache for AI leaders.

“That went from, at the beginning of this year, an issue that never came up — people were totally happy with the amount they were spending — to all of a sudden, a huge issue,” OpenAI CEO Altman admitted during an event last week.

Duh

“I think we’ll have a lot of ways we can help people get more value for less spend,” he added.

A forlorn hope applies. Tom Wolfe's The Bonfire of the Vanities come to mind here.

The Bonfire of the Vanities, is a 1987 novel by Tom Wolfe. The story is a drama about ambition, racism, social class, politics, and greed in 1980s New York City, and centers on three main characters: WASP bond trader Sherman McCoy, Jewish assistant district attorney Larry Kramer, and British expatriate journalist Peter Fallow. 

Works in 2026 as well, right?

It’s a major conundrum for all players involved. AI companies have been bleeding tens of billions of dollars as costs for data center construction projects mount. Cutting prices now could make the situation even more dire, deteriorating already disastrous profit margins.

Anthropic and OpenAI have been caught in a heated race, with the former making major gains through its enterprise-focused coding tools as of late. Its recent advancements clearly rattled the latter, considering the latest news.

Both companies have confidentially filed for an IPO within the last ten days, raising enormous stakes. However, the fact that both are scaring away new users thanks to soaring prices isn’t exactly a vote of confidence to investors, which could soon force AI executives to rethink their business models.

I'm glad they have confidence but ...

Illustration by Tag Hartman-Simkins / Futurism. Source: Getty Images

Companies that fell head over heels for AI are experiencing a rude awakening.

Costs to access powerful AI tools are soaring, forcing company leaders to ask some difficult questions. As Axios reports, the early warning signs are already here, with Microsoft planning to remove its Anthropic Claude Code licenses after opening up access to the tool just six months ago, reportedly for financial reasons.

Seems the 95% failure rate for AI pilots, as per MIT holds.

After reading all of this, would you buy into their IPOs? I thought so. 

Thinking Betelgeuse

Betelgeuse is a lumpy clump of boiling gas.
(Image credit: ALMA (ESO/NAOJ/NRAO)/E. O’Gorman/P. Kervella)

 Betelgeuse is getting ready to blow and scientists and yours truly can't wait to see it happen.

Orion constellation map

When Betelgeuse explodes in a supernova, it will shine as bright as the full moon in our sky.

Located some 650 light-years from Earth, Betelgeuse, also known as Alpha Orionis, usually ranks as the tenth brightest star in the night sky. The star, however, is known for its periodic dimming and brightening up, which sometimes makes it climb a bit up or down the ranking.

Astronomers think that Betelgeuse is only 10 million years old, that's 50 times younger than our sun. Despite its relatively young age, Betelgeuse has already run out of the hydrogen fuel in its core and is now in the final, red giant stage of its life, fusing helium to carbon. 

The reason for Betelgeuse's fast-paced life is the fact that it was likely born very massive. According to ScienceAlert, Betelgeuse used to be a blue-white O-type star, the most massive kind of main sequence, hydrogen-burning star. The bigger the star, the brighter it shines and the hotter it burns. But also the faster it runs out of its hydrogen and the sooner it turns into a red giant. 

In its prime, Betelgeuse must have been tens of times more massive than our sun and tens to hundreds of times more luminous. Temperatures on its surface may have reached up to a mind-boggling 89,500 degrees Fahrenheit (50,000 degrees Celsius), compared to the sun's "lukewarm" 10,000 degrees F (5,500 degrees C).

Which means, like octopuses, Betelgeuse's lifetime is short but intense and now it's time is almost up,
which relates indirectly to this ...

The fragility of  SpaceX's products stagger the imagination. Rockets are "controlled" exploding bombs, extremely fragile, prone to blowing up and very touchy to maintain. The variables regarding the Starship and it's booster, with it's 32 engines, gives one pause to the max. This is akin to data centers and silicon as pushing electrons through chips and networks generate intense heat and requires both extreme power and cooling to keep the tech running and we haven't even discussed the problems of maintenance and replacement of hardware to keep said appliances online to service the needs of hundreds of millions of customers 24/7.

Compare both to the Space Elevator and Microtubuels to see why both of theses enterprises are, IMHO, but a short term solution, something the bond boys are beginning to realize with great trepidation. 

No silver bullet

Why this the book again??? It's been talked about ad nauseam in BRT. Adding late people to a late project makes it later, the plan often takes longer than the execution. blah, blah blah but there's a reason. The reason is because AI has the workflow ass backwards. It's not the code that's the problem, it's the workflow as AI now has all the content needed but doesn't have a workflow to leverage the code. For starters, lets talk problem solving, areas of interest and encouraging the user (I hate that term) in terms of what's needed. Thinking things through applies.

What this writer would like to see in terms of working with AI.

Give users choices. Just have the existing text bar in place with a list of clickable topics below. Arrange them by category and let the person begin the journey by selecting the category. Keep it spare. Think early Apple in terms of interface. Limit drill downs. Don't make me think - Jobs, applies.

Click here for the 1992 apple manual.

Using the current model, AI front loads everything. Nature doesn't work that way because nature's lazy because the principle of least action applies. Why do the bit AI companies violate that one essential? 

Balancing a pendulum. By moving the little wagon to compensate for the instability of the pendulum,
the pendulum can remain vertical. :) This is akin to a seal moving his head  around in
balancing a broom stick on its nose.

Some AI companies are finally understanding this by having the AI be a partner and not a tool. Assume nothing. When looking at the task bar and seeing What can I do for you today? forces the user to ground 0. It's a waste of time and a waste of compute. No names here but one outfit is doing Fred Brooks. The user and AI work together to solve a problem but only when the users selects a specific problem to solve in relationship to the target discipline before the AI does anything. This is common sense.

Real world problem. My client had a problem. Link graphics to packaging and to the item that goes into the package. Me: What's the biggest pain in the ass you have in dealing with the current system? Ask without judgement. Stand back. If the people in the company trust you, you will find out in a nanosecond where the problem lies. 

As per von Clausewitz, this, for the client, is the center of gravity. This is the crux of Fred's TMM. You have to understand the problem first. You have to ask questions as the user will know more about the problem then you ever will. Current AI companies don't do that. It's big tech, we know because AI solves every problem known to man and we'll tell you how to do it. 

Brevet Major General George Armstrong Custer in field uniform.

Ask Custer about the Little Big Horn as he knew everything, right? Know the ground. The Art of War.

This is why 95% of all AI projects fail, both on the part of the client companies involved and on the part of AI companies also because there is no silver bullet in solving complex problems, no easy way out for anyone and the AI industry is finally beginning to learn why this take rings true.

Remember: Complexity is easy, simplicity is hard. Keep things simple, not simpler. Einstein.

The IPO ...

Photograph: Chris Ratcliffe/Getty Images

IPOs do weird things to people and organizations as greed, hype and the tell combine
to see just how many investors can we get to buy into "the next big thing."

IPO PR trick one, deny competitors to the family jewels

Anthropic is backtracking on a policy that would have covertly limited competitors from using its new AI model, Claude Fable 5, to develop other AI models. The company changed course after the move received significant backlash from the AI research community.

“We’re changing Fable 5’s safeguards for frontier LLM development to make them visible,” Anthropic said in a statement to WIRED. “We made the wrong trade-off and we apologize for not getting the balance right.” Anthropic released Claude Fable 5, a version of its latest AI model with additional safety guardrails designed to prevent misuse, earlier this week. Some of the safeguards Anthropic decided on were unsurprising: The company said it would reroute users who asked questions about cybersecurity, biology, or chemistry to a less capable AI model to reduce the chances of someone using the advanced AI to carry out a cyberattack or build a bioweapon.

But for researchers trying to use Claude Fable 5 for frontier AI development, Anthropic outlined a different approach. The firm would deliberately degrade the model’s performance in ways that were invisible to the user. The move would effectively sabotage researchers trying to use Claude to train competing AI models, which Anthropic explicitly bans in its terms of service.

Anthropic now says it’s changing course, and that Claude Fable 5’s safeguards for AI development will be visible to users. If the company suspects a user is trying to use Claude to build a highly capable AI, it will alert them that it’s either refusing the request or rerouting the user to a less capable model.

IPO PR trick two, the self evolving AI. Conway's Game of Life comes to mind here. :)

Taken far enough, and given enough compute, that trend points to an AI system capable of fully autonomously designing and developing its own successor. This is called recursive self-improvement. We are not there yet, and recursive self-improvement is not inevitable. But it could come sooner than most institutions are prepared for.

AI that can build itself would be a major development in the history of technology—one that could bring enormous good for the world in science, healthcare, and beyond. But full recursive self-improvement also might increase the risks of humans losing control over AI systems. If systems are capable of fully building their own successors, the ways we secure them, monitor them, and shape their behavior all grow much more important.

In English, what are we really talking about? Bloatware, lay code on top of code, requiring ever more compute but more ominously, it's inbreeding 101, the ouroboros eating its tail, a perfect way to hallucinate in new and better ways but you, my loyal investors .. we need ever more capital to makes this grand vision a reality, hence the upcoming IPO for the ages.

Wall Street analysts and financial forecasters widely expect the initial public offering (IPO)
to officially price and list around October 2026

A dragon-like ouroboros in a 1478 drawing in an alchemical tract[1]

Exuberance 101

Illustration by Alicia Tatone. Sources: Brett Carlsen / Getty; iStock / Getty.

When watching Pentagon Pete, for the first time, talking pushups to flag officers and senior ems, yours truly was seeing a guy completely lacking in self confidence as the bravado, the wearing of suits two sizes too small and taking on a swagger of no substance, all in the attempt of being in total command of the situation, collapsed in a nanosecond as he's not a serious person but I digress. The Atlantic has it covered.

The defense of the United States is a serious business. Every day, men and women, civilian and military, attend to the smallest details—the caloric content of a soldier’s meal, the fabric in a uniform—while others advise senior leaders on the use of violence to achieve the ends of national policy. Some, stationed underground, underwater, or on bomber airfields, stand ready to fulfill orders with apocalyptic consequences. These people are professionals and carry themselves with the quiet pride that comes from serving their nation.

And then there’s Pete Hegseth, the self-declared secretary of war.

Hegseth approaches his job as if it’s a vacation rather than public service. He shows up on military bases looking as exuberant as a bored househusband who just got a kitchen pass from his wife for a weekend of paintball with the boys. He sports a pocket handkerchief that looks like an American flag. He carries himself with an effortful swagger that is meant to convey machismo but instead just looks awkward. He talks with practiced drama, like he’s constantly trying out for the lead in a school play.

He is, for want of a better word, corny. Corniness isn’t always bad; sometimes it comes from people who are so effusively sentimental or expressive that others laugh at them, albeit with a certain amount of indulgence. But Hegseth’s behavior is not the endearing corniness that comes spontaneously from an overly earnest person trying to express great emotion. It is the overbearing corniness that comes from trying to mimic deep sincerity, and it tends to end up sounding like a cross between a late-night-television preacher and an arrogant luxury-car salesman: Jesus brought you here, my brother, so what’s it gonna take for you to fly home in one of these super-lethal F-35 babies today?

Hegseth’s public statements are full of such corniness. This is someone who thought that changing the name of the Department of Defense to the “Department of War” would convey strength. Perhaps a better man could have sold the idea that President Harry Truman made a mistake changing the name during the Cold War, but Hegseth is not that man. “Maximum lethality, not tepid legality,” he semi-rapped at the time, “violent effect, not politically correct.” This political doggerel is meaningless—and because he seems to have practiced before saying it, it is also cringe-inducingly corny.

The mighty Corn Dog comes to mind without question. :)

A hand-dipped corn dog on a stick, served from a traveling food vendor trailer
at Rochesterfest in Rochester, Minnesota

A corn dog (also spelled corndog and also known by several other names) is a hot dog sausage on a stick that has been coated in a thick layer of cornmeal batter and deep fried. It originated in the United States and is commonly found in American cuisine.

Sound delicious does it not?

I bet ...

Image credit: VentureBeat with ChatGPT

Seems a tiny entity, Sapient, an outfit with connects to China, Singapore and California, has emulated The Mythical Man Month in terms of AI and coding by separating instructions from calc, thus reducing bloat and inefficiencies to the extreme while costing $1500 to build. Better yet, it's open source as they believe in free as in beer. The code is free but you pay for services. :)

Try their HRM code on Github for free.

Training a foundation LLM from scratch costs millions and requires internet-scale data — which is why most enterprises don't bother. Sapient thinks it has a cheaper path.

To overcome this brute-force scaling dogma, researchers at Sapient developed HRM-Text, which replaces standard Transformers with a highly sample-efficient Hierarchical Recurrent Model (HRM), an architecture they first introduced last year.

HRM decouples computation into slow-evolving strategic and fast-evolving execution layers. Instead of brute-force autoregressive prediction on raw text, HRM-Text trains exclusively on instruction-response pairs. This is close to real-world enterprise settings, where users usually expect a targeted answer to a specific task.

The researchers were able to train a 1B-parameter HRM-Text from scratch at a fraction of the cost and tokens of normal LLMs. Their model achieved performance competitive with much larger open models on key industry benchmarks.

For real-world AI applications, this means foundational pretraining is no longer restricted to highly resourced institutions. With HRM-Text, organizations can affordably pretrain their own highly capable reasoning models from scratch and pair them with external knowledge stores.

I bet the bond holders are thinking really hard about all the billions of dollars spent
to build AI and data centers that are eating the world as we speak. I know I would.