Begin Before Anyone Believes – Be It Deep Tech or Mountaineering
From death zones to Starcloud's orbital leap: this is the pattern of conviction.
// 4–5 min read · Two seemingly impossible stories. One timeless pattern. One crew summited 14 death zones in record time. Another is launching the future into orbit.
Disclaimer: I’m not a space expert. I didn’t grow up building satellites or optimizing heat dissipation at zero gravity. I just read obsessively and do Snoopy moves when a topic hooks me.
Data centers in space?
At first, it sounded like sci-fi. Then I started digging. Turns out, it’s one of the clearest bets in deep tech if you understand what’s coming in AI, energy and infrastructure.
The Essentials Up Front:
Core insight: Real benchmark isn't what's been done or what’s bound to Earth. It’s what becomes possible when you move with conviction. Within your power.
AI training and inference demands are outpacing Earth's computing infrastructure. Starcloud skips the bottlenecks by leaving Earth as the math doesn't work on the current trajectory.
Starcloud is building orbital data centers that generate energy 22x cheaper than Earth-based alternatives with unlimited scale, built-in cooling. Not a thought experiment. A 2027 launch window is locked in.
Visionary founders don’t wait for consensus. They move before it sounds reasonable. That’s how the future gets built.
Next Up: Launching our AMA series with founder Philip Johnston, followed by their angel investor. Watch this orbit.
Some feats don’t wait for consensus. Nimsdai climbed all 14 death-zone peaks in under 7 months. No one believed he could. Starcloud’s making the same kind of bet. This time, above Earth.
“First tell yourself what kind of person you want to be, then do what you have to do. For in nearly every pursuit we see this to be the case. Those in athletic pursuit first choose the sport they want, and then do that work.”
— Epictetus
00/ Project Possible
In 2019, a former Gurkha soldier named Nimsdai Purja set out to do the impossible.
No one in the mountaineering world had heard of him. He announced: "The fastest time to climb all fourteen 8,000-meter peaks was less than 8 years. If I can stay alive, I can do this in 7 months."
The entire climbing world responded: "Who the heck is Nimsdai Purja?"
To put this in context: these are the only summits higher than 8,000 meters in the notorious “death zone” where human life cannot exist. The first successful attempt took 16 years. Normally, a single eight-thousander is a career-defining endeavor.
Almost no one believed in him. Mom was sick, family had financial problems. Nims didn't have Red Bull money or backing from the big alpine clubs.
What he had was grit, an audacious dream and a sense of purpose for his country. Also, apparently, superhuman lung capacity.
He called it Project Possible.
And he did it in 6 months and 6 days.
The real benchmark isn't what's been done. It is what is possible when you stop looking at others and start moving with conviction. Within your own power.
Five years later, another group of unknowns would face their own "Who the heck are these guys?" moment. Not on Earth's highest peaks, but in the space between the world and the sun.
I have been watching them since they first appeared on Y Combinator's company directory.
01/ The Math That Breaks in 2027
Here's the mismatch nobody wants to talk about.
We’ve entered the most aggressive phase of techno-capital mobilization in decades. A new industrial revolution is underway and it's powered by AI inference. We are not just training smarter models, we are building a new planetary stack of AI infrastructure, one GPU, one gigawatt at a time.
The timeline problem is simple arithmetic: Training next-generation models starts in 2027 (24 months away), but AI energy consumption is growing faster than Earth's power infrastructure can handle. Getting permits for new power infrastructure takes a decade. You don't need a PhD to see this doesn't add up.
Demand for power for data centers is expected to rise significantly in the United States.
Back in late 2023 (still valid), McKinsey projected that US data center energy consumption would quadruple by 2030. That’s the kind of energy math no one’s ready for yet Starcloud is building for.
The quotes tell the story:
Sam Altman loud and clear: "We still don't appreciate the energy needs of this technology...there's no way to get there without a breakthrough."
Elon Musk sees the cascade coming: "We have silicon shortage today, a voltage step down transformer shortage probably in about a year, and then just electricity shortages in general in about two years."
Mark Zuckerberg agrees: "We would build out bigger clusters than we currently can if we could get the energy to do it."
Tom Mueller from SpaceX puts it starkly: "The amount of power to run compute by 2045 will be the base power of the planet right now."
These aren't distant warnings. The decisions for 2027 are being made right now by people who clearly see the real bottleneck. In other words, the world's smartest people are all saying the same thing: “Houston, we have a power problem.”
In short: AI is about to outgrow Earth. Turns out, when you need the power of entire nations to teach computers, maybe it's time to look up.
Enter the builders who decided to skip Earth's limitations entirely.
02/ Meet Starcloud (née Lumen Orbit)
Where's the best place to build a data center? While most people hunt for coordinates on Earth, the daring answer is "Not on Earth at all."
They are building megawatt-scale data centers in space, using satellites in low Earth orbit to process data.
Founded in January 2024, the founders closed a $21 million seed round in mere weeks.
The investor list tells a compelling story:
NFX as lead investor
Y Combinator
FUSE
Soma Capital
Scout funds from Andreessen Horowitz
Scout funds from Sequoia Capital
Cloud AI Accelerator
NVIDIA Inception Program
Over 200 VCs fought to get in. Turns out 200 VCs suddenly became space enthusiasts overnight. This reflects broader deep tech investment trends as smart money moves toward infrastructure-scale opportunities.
Why Redmond, Washington?
They chose to build in Redmond for one clear reason: it's the global satellite manufacturing hub. SpaceX's Starlink and Amazon's Project Kuiper are based there. Redmond companies have built over half of all satellites launched worldwide, and with mega-constellations coming, that could soon top 75%. When you are hiring engineers to build data centers in space, geography matters.
The tech world's response mirrored the climbing community five years earlier: "Who are these guys?"
Fair question. Meet the trio who decided Earth's infrastructure bottlenecks were optional.
Philip Johnston | Co-founder &CEO :
Former McKinsey consultant who advised Middle Eastern governments on satellites. Previously founded Opontia, raised $62 million total, scaled to 100 people, sold it. Beyond the stellar resume, he has a give-back mentality that shines through his actions.
Fun fact: Philip has an identical twin, Adrian Johnston. Both are YC-backed founders building ambitious companies on separate tracks. Must be something in the DNA.
Ezra Feilden | Co-founder &CTO:
Decade at Airbus Defense and Oxford Space Systems where he worked on satellite design, specializing in deployable solar arrays and large deployable structures. One of the rare minds who got to work on NASA's Lunar Pathfinder. He's been thinking about space infrastructure longer than most people have been thinking about AI scaling.
Adi Oltean | Co-founder &Chief Engineer:
Ex-SpaceX engineer who had independently developed the same orbital data center concept. When Johnston and Feilden got the idea, someone introduced them. He joined within a month.
In his previous startup, he engineered unique solutions for in-orbit assembly and manufacturing of large structures in space. Before that, he spent 20 years at Microsoft working on large production GPU clusters and putting his name on more than 25 unique patents.
The Team's Conviction
The team’s conviction comes from seeing what's actually happening at SpaceX. Philip visits their Starbase facility regularly, each time, the Starship Gigafactory is 10 times bigger than before. Three Starships per week within 18 months, then three per day. That's 1,000 times more payload capacity, not just cheaper costs.
They are not just building access to space. They are mass-manufacturing it.
Here's what separates them from most space startups: speed. Most space companies take 5+ years to launch. Starcloud is targeting 18 months from founding to orbit. A 2027 launch window is locked in.
What most startups call a moonshot, Starcloud calls a launch calendar.
🎬 If Starcloud were a movie…
Philip Johnston would be the lead in a Nolan film. Clean-cut, mission-first, probably delivering the line “The clock’s already ticking” while a rocket launches in the background. He’s not just the polished operator with Ivy League degrees. He’s the calm-eyed optimist quietly outthinking everyone in the room.
Ezra Feilden is the CTO who sketches deployable solar arrays on napkins at lunch, but makes you feel like you are helping build something sacred. His engineering isn’t just technical, it’s devotional.
Adi Oltean is the quiet force behind the curtain. An ex-SpaceX engineer who’s seen rockets fail and still bets on the stars. He doesn’t speak in slogans. He speaks in blueprints. And when he nods, you know it’s going to work.
03/ When Physics Trumps Skepticism
Here's where the impossible becomes inevitable.
Energy Advantage: Orbital computing infrastructure powered by space-based solar arrays achieve 95% capacity factor vs 24% for Earth's best solar farms. In orbit, panels generate roughly 5 times more energy due to continuous sunlight and no atmospheric losses.
Cost Inversion: Operating a 40MW data center for 10 years costs $140 million in electricity on Earth. In space? $10 million for launch plus solar arrays. That's 22 times cheaper. (These figures depend on optimistic launch costs.)
Scale Physics: A 5-gigawatt cluster needs a 4km x 4km solar array in space, feasible with modular design, nearly impossible to acquire on Earth.
Cooling: Deep space averages -270°C. No energy-intensive chillers needed, just deployable radiators using vacuum as a heat sink. Though scaling radiators to gigawatt levels remains a significant engineering challenge.
Here's the scale of that challenge: the largest heat dissipator ever deployed in space is the size of a tennis court. Starcloud is planning one that's 1 kilometer by 4 kilometers. As Philip puts it: "Nothing against the laws of physics, just a very large engineering challenge." Sometimes the biggest leaps happen when someone treats "impossible scale" as an engineering problem, not a physics problem.
Third-party validation: The European Commission's ASCEND study and Thales analysis confirm orbital data centers could be more eco-friendly than terrestrial alternatives. Google and Microsoft are reportedly exploring similar concepts.
The Honest Moment:
Skeptics often raise concerns about radiation shielding, collision avoidance, and thermal management in low Earth orbit. All valid points. These aren't trivial problems.
However, SpaceX’s Starlink constellation demonstrates that these challenges can be effectively managed: the satellites operate at similar altitudes using silicon-based solar panels with minimal shielding, relying on robust engineering and operational strategies rather than heavy, traditional protections.
But when the physics make this much sense, the question isn't if. It's who gets there first.
Here's what most people miss: the biggest risk isn't technical failure, it's being too late. While everyone debates orbital mechanics, the launch window for market leadership is shrinking faster than the technical challenges.
The Energy Curve Nobody Wants to See
As AI’s appetite for power accelerates, it’s becoming painfully clear that Earth’s infrastructure isn’t ready. The chart below shows the trajectory we’re on unless we change course or altitude.
04/ Why Smart Money Is Already Moving
The pattern is always the same: first the physics, then the validation, then the gold rush.
We are somewhere between validation and gold rush.
Industry insiders see it coming. Tom Mueller, SpaceX employee #1, couldn’t be any clearer: "The amount of power to run compute by 2045 will be the base power of the planet right now. The drain on resources is so high, you need to put that compute in space and use the power of the sun."
The convergence of trends is obvious. Plummeting launch costs meet exponential AI demand meet energy bottlenecks. When three exponential curves intersect like this, someone's about to get very rich or very broke.
While others debate endlessly in PowerPoint, they are actually building things. 18 months from founding to orbit vs the industry standard of 5+ years. As Johnston says and I 100% agree :
"We'd rather fail fast in space than succeed slowly in conference rooms."
Apparently, the team finds it easier to build data centers in space than PowerPoint slide decks. Can you blame them?
Everyone loves ambition in retrospect.
While it's happening, it looks insane. The question is: are you comfortable being early or do you prefer being obvious?
The infrastructure of intelligence is leaving Earth's atmosphere while most people are still debating Earth-bound solutions.
05 / The Decision Framework That Matters Now
Here's the real question hiding behind all the technical specs and funding rounds: Are you making decisions based on current constraints or emerging possibilities?
The meta-insight here isn't about space or data centers. It's about conviction timing. History shows us one thing repeatedly: the future has never been built on the obvious. Because the moment it becomes obvious, it's not the future anymore. It's right now.
Starcloud chose to be the kind of company that acts on physics rather than permissions. The kind that builds solutions rather than presentations.
Via Negativa
Their core mindset? Via negativa. Success by subtraction.
Instead of solving Earth's energy bottlenecks, they eliminated Earth entirely.
Instead of navigating decade-long permitting processes, they chose space where permits don't exist.
Instead of building bigger cooling systems, they moved to where cooling is built-in.
Sometimes the smartest solution isn't adding more—it's removing the constraint entirely.
If I were to build my own startup today, I would ask myself these three questions after reflecting on this piece:
Do I believe this enough to go all in before anyone else does?
Check point: Conviction doesn’t ask for consensus. It acts before it arrives.
Is this the window before it sounds obvious?
Timing isn’t luck. It’s knowing when to act before the crowd.
Am I building on a terrain that scales?
A good idea in the wrong infrastructure dies quietly. A good idea also needs healthy balance sheets. You need healthy unit economics and meaningful impact both, not either.
In life and tech, we’re all benchmarked by the boldness of our decisions, not the safety of our timing.
06 / The Questions Only Philip Can Answer
You've read the physics. You've seen the validation. Hopefully you felt the conviction between the lines.
But here's what you can't get from white papers or press releases: the real story from someone actually building this.
I'm launching a new AMA series under Essentialist Edge. Kitchen-table conversations with contrarian minds. Campus vibes.
Part Lex Fridman’s quiet intensity, part Rogan’s curious chaos. Unscripted, unpolished and maybe slightly unhinged. Always signal.
Not just AMA. It’s AM-WHY & HOW.
And my first guest is... drumroll... Philip Johnston. You’ve probably guessed by now.
Not a podcast. Not a panel. Just real conversations with people who love tech for what it does for humanity, not the hype.
We’ll record selectively but the focus is real time. Honest. Unscripted.
Next AMA? Their angel investor. We’ll flip the lens.
Why now?
Deep tech has long been treated like the awkward genius in the corner. Brilliant, misunderstood. Now it’s finally having its moment. And it deserves real, global conversations.
Interested? Reply to this newsletter or DM me on Substack.
07 / The Deeper Game
Everyone else sees impossible. They see inevitable.
Nims climbed 14 death-zone peaks in 6 months. Philip, Ezra and Adi are building AI infrastructure in orbit. Same conviction. Different altitude.
The next era of artificial intelligence infrastructure won't be powered on Earth. It's being built as space-based infrastructure where energy, scale and cooling aren't bottlenecks above the clouds, beyond doubt.
Want to know how big they are really thinking? The team's ultimate vision is a Matryoshka brain, a Dyson sphere connected to as much compute as a star system can physically support. Suddenly, orbital data centers look like step one of a million-year plan.
The question isn't whether this will happen. It's whether you'll be among those who saw it coming. As for Starcloud, time will tell where this will lead them.
Courage whispers… Leave base camp, leave convention, leave what’s already been done.
Because the future doesn’t wait at base camps. Neither should you.
End of Line.
{ Next release loading... }
Stay essential,
Nihal
Reminds me of the quote "The reasonable man adapts himself to the world; the unreasonable one persists in trying to adapt the world to himself. Therefore all progress depends on the unreasonable man”
They took it one step further, they went off world!