No Disassemble 5: The Shovel Apocalypse

Or: How the AI Gold Rush Solved Everything and Created New Problems Faster Than We Could Invent Them

April 1, 2027

Before we begin, I should probably disclose a conflict of interest.

For several years I fed almonds to the local squirrels.

Not occasionally.

Systematically.

There was a squirrel route. There were regulars. There may have been expectations.

At some point my wife purchased a package of tiny unicorn hats designed specifically for squirrels.

https://a.co/d/0i9E4tAS

I would like the record to show that this was her idea.

I would also like the record to show that I immediately participated.

Several squirrels tolerated the hats.

One appeared delighted.

Another seemed to believe he had been chosen to fulfill an ancient prophecy.

At the time I believed this was harmless fun.

In hindsight I may have been interfering in local squirrel politics.

This detail will become important later.

It is difficult to know for certain because squirrels are notoriously bad at filling out customer satisfaction surveys.

This detail may seem unrelated.

It is not.

Like most things in this story, it is connected to everything else.

At the time I believed I was feeding squirrels.

Years later I would discover I had accidentally become involved in water policy, artificial intelligence, semiconductor manufacturing, data center infrastructure, zoning board politics, watershed management, and an ongoing dispute with several squirrels regarding the ethical implications of almonds.

The dispute escalated.

I would like to stress that none of this was intentional.

The whole thing started because I wanted to know how much water AI uses.

That seemed like a simple question.

It was not.

Somewhere around 2023 a few prospectors wandered into the digital wilderness, kicked a strange glowing rock, and discovered that if you stacked enough GPUs together, the pile would begin answering questions.

The response from civilization was measured and reasonable.

Everyone immediately lost their minds.

The newspapers called it Artificial Intelligence.

Wall Street called it Growth.

The venture capitalists called it Opportunity.

The tech bros called it Destiny.

The artists called it theft.

The environmentalists called it a water problem.

The philosophers called it a crisis.

The governments called it strategic infrastructure.

The lawyers called it billable hours.

The data centers called it Tuesday.

Within months every major corporation in the developed world was sprinting through the hills carrying digital gold pans and screaming:

"There's intelligence in them thar servers!"

Nobody knew exactly what they had found. Nobody knew exactly what it would do. Nobody knew exactly what it was worth.

These uncertainties did not slow anyone down.

In fact, they seemed to increase enthusiasm.

Soon every shovel factory in civilization was operating at maximum capacity.

"What in tarnation?" asked ordinary citizens. "I just wanted an SD card."

Unfortunately they had unknowingly wandered into a gold rush. And in a gold rush, the people who get rich are rarely the miners.

The people who get rich sell shovels.

The GPU companies sold shovels. The cloud companies sold shovels. The power companies sold shovels. The networking companies sold shovels. The cooling companies sold shovels. The construction companies sold shovels. The consulting companies sold maps explaining where the shovels were located. Other consulting companies sold reports explaining why the first maps were insufficiently shovel-aware.

Several startups raised funding to create AI-powered shovel deployment optimization platforms for next-generation excavation ecosystems.

Nobody knew what that meant.

Investors approved it anyway.

Soon there were not enough shovels. This was considered unacceptable. A bipartisan commission was immediately formed to determine why there were not enough shovels. After eighteen months and six hundred pages of findings, the commission concluded that additional shovels would be required.

This recommendation was regarded as groundbreaking. The report was praised for its depth.

New shovel factories were built. This solved the shovel shortage. Unfortunately it created a shovel operator shortage. Universities responded by launching shovel science programs. Within two years there were more shovel scientists than shovels.

This was widely celebrated as a workforce development success.

By 2026 AI systems had become capable of designing superior shovels. The first AI-designed shovel was twenty-three percent more efficient. The second AI-designed shovel was forty-seven percent more efficient. The third AI-designed shovel successfully eliminated the need for digging entirely.

Nobody understood how it worked.

This was not considered a barrier to investment.

The company received three billion dollars in funding.

Wall Street discovered shovel derivatives. The value of the global shovel market exploded. The number of actual shovels remained largely unchanged.

The economy was now mostly shovels studying other shovels.

Somewhere beneath the paperwork there was still a hole somebody had originally intended to dig.

As demand exploded, strange things began happening. The cost of memory cards rose. The cost of electricity rose. The cost of land near substations rose. The cost of water rights rose. The cost of explaining any of this to your relatives during Thanksgiving rose dramatically.

That was the point I got involved.

Not because I cared about shovels. Not because I cared about gold. Not because I cared about venture capital.

I wanted to know how much water AI uses.

That seemed like a simple question.

I was wrong.

The first answer led to data centers. The data centers led to electricity. The electricity led to cooling systems. The cooling systems led to water. The water led to aquifers. The aquifers led to watersheds. The watersheds led to zoning boards. The zoning boards led to permitting. The permitting led to governance. The governance led to history. The history led to people who had already spent thousands of years living inside versions of the same problem.

The history also led to me reading county planning documents at two in the morning and wondering where I had made a wrong turn.

The question those people had been asking was not: "How do we maximize extraction?" Not: "How do we optimize shareholder value?" But: "How do we keep the watershed alive?"

Ten thousand years is not proof that somebody was right about everything.

But it is long enough that maybe you should at least listen.

Especially before building another shovel factory.

That was the moment I realized the water question was never actually about water. Water was just the doorway. The real question was: who gets to decide what the water is for?

Then the squirrels got involved.

The squirrels were furious.

Not because they hated AI. Not because they hated miners. Not because they hated data centers.

The squirrels were furious because the trees were thirsty.

This was a surprisingly practical concern.

Unfortunately the squirrels launched their campaign online. The memes went viral. The posts spread. The videos exploded. The miners were denounced. The planners were denounced. The shovel companies were denounced. Several squirrels denounced other squirrels for insufficient denunciation.

Engagement reached historic levels.

Mining activity increased.

"What in tarnation?" asked the squirrels.

Analysts later determined that each viral anti-mining meme required substantial shovel-powered infrastructure. The squirrels responded by creating additional memes explaining why this statistic was misleading. These memes also required infrastructure.

This was not the outcome they had intended.

At this point I felt I finally understood the system. I had investigated the miners. The planners. The economists. The environmentalists. The data centers. The squirrels. The watersheds. The whole thing.

I sat down to write my conclusion.

The future, I wrote, belongs to those capable of balancing competing optimization functions while maintaining long-term systemic integrity across interconnected stakeholder—

Tap.

Tap.

Tap.

A squirrel sat outside my window. A very small squirrel. Holding an acorn. Looking directly at me.

"Well aren't you a cute little guy."

The squirrel tilted its head.

"Oh, what a cute little guy you ahghhgh—"

The front door opened.

Several hundred squirrels had apparently figured out the lock.

The rest of the evening was educational.

The squirrel leader climbed onto my desk. I recognized him immediately. He was wearing one of the unicorn hats. It had been modified with a pirate emblem.

Apparently local squirrel politics had evolved considerably since our last interaction.

"I was on your side!" I shouted.

The squirrel leader adjusted his tiny pirate hat.

"I know."

"Then why are you doing this?"

"Because that's not a side."

"What?"

"That's a position."

Several squirrels nodded. One appeared to be taking notes. Another was livestreaming. A third was eating almonds that, in retrospect, I may have provided.

The squirrel leader looked around the room.

"You spent a year investigating miners."

"Yes."

"You investigated data centers."

"Yes."

"You investigated water."

"Yes."

"You investigated us."

"Also yes."

"And what did you learn?"

I thought about it. The aquifers. The zoning boards. The permits. The data centers. The watershed. The trees. The water. The squirrels. The whole impossible interconnected mess.

Before I could answer, a younger squirrel stepped forward.

He was carrying a chart.

I immediately became nervous. Nobody likes where a story is going when the squirrels arrive with data.

The younger squirrel placed the chart on my desk.

"According to this," he said, "one almond uses dramatically more water than a peanut."

The room fell silent.

The squirrel leader stared at the chart. Then at the almonds. Then back at the chart. Then at the almonds again.

"I don't like this data," he said.

"Neither do I."

"Are the numbers correct?"

"Unfortunately."

"Can we denounce them?"

"You can try."

"Will that change them?"

"No."

"This is the worst kind of data."

Another squirrel stepped forward.

"What about peanuts?"

Several squirrels gasped. One fainted.

The squirrel leader slowly turned.

"Peanuts?"

"According to the chart, peanuts require substantially less water."

"We're not discussing peanuts."

"Why not?"

"Because we're angry."

"Yes, but we're angry about water."

"Correct."

"And peanuts use less water."

"Also correct."

"Then shouldn't we eat peanuts?"

The room became uncomfortably quiet. For the first time all evening nobody was yelling. Nobody was livestreaming. Nobody was denouncing anybody.

The squirrel leader looked exhausted.

"Oh no."

"What?"

"The numbers followed us home."

A long silence filled the room.

Finally the squirrel leader removed his tiny pirate hat.

"Fine."

"Fine?"

"Fine. We're switching to peanuts."

"Just like that?"

"No. There will be months of arguments. Three position papers. Two podcasts. A schism. Several angry influencers. At least one squirrel will claim peanuts are a corporate psyop."

"That sounds about right."

"But eventually?"

The squirrel leader sighed.

"Eventually."

The younger squirrel nodded. The charts were collected. The almonds were confiscated.

The room slowly emptied.

As he reached the door, the squirrel leader turned back.

"Kevin."

I looked up.

"Yes?"

"The moment you show us data proving sunflower seeds are better than peanuts, this starts all over again."

Then they left.

Somewhere in the distance another shovel factory was approved.

The miners celebrated. The planners celebrated. The economists celebrated. The AI companies celebrated. The environmentalists celebrated. The streamers celebrated. The gamers celebrated. The peanut industry celebrated. The almonds did not celebrate.

Nobody was entirely sure why.

Reality, meanwhile, remained undefeated.

The charts were archived. The arguments continued. The aquifer remained unimpressed.

The planning commission meets again on Monday.

The National Strategic Shovel Reserve declined to comment. A bipartisan commission has been formed to investigate the matter. Its preliminary findings suggest that additional investigation will be required. This recommendation has also been widely regarded as groundbreaking.