Real-time signal from X with AI Insights, Price targets & investment scores on every ticker. and AI recommendation. On Web, iPhone & Android.
San Francisco, CA
Joined December 2020
- Tweets18.7K
- Following332
- Followers7K
- Likes690
Philip Johnston says Orbital AI data centers make sense below about $500 per kilo launch.
“One of the big biggest topics and I know we talked about this with Adi is the economics of this. You are making more economical satellites going out in space and the vehicle itself and so that's one thing and so I'd love to break that down in the economics there and then also further drill down the difference between on land like terrestrial data centers versus orbital.
Yeah so the whole thing…”
@sourceryy @mollysoshea
Watch the full pod:
BREAKING: Inside Starcloud, the $2.3B Startup Putting Data Centers in Space
FACTORY TOUR - Redmond, WA
Primes quoted them $75M-$100M to build their first data center in space. They did it for $2.5M.
In just 21 months, Starcloud is one of the fastest companies to go from founding to orbit.
With $450M in total funding from Manhattan West, Nvidia, Cisco, Benchmark & EQT, the team walks me through their mission for AI in space.
I toured the new HQ w/ CEO @PhilipJohnston + co-founders @EzraFeilden & @AdiOltean
AI compute on Earth is limited by power & permitting. @Starcloud_ is moving it to orbit, where solar power is nearly constant. Philip says the dawn-dusk orbit alone could fit about 10 TW of compute, roughly 20x the entire US power grid. Once launch costs fall below $500/kg, he says compute in space will be cheaper than on the ground.
From Starcloud 1, Starcloud 2, to Starcloud 3, we saw:
Tracking Starcloud 1 live in orbit, circling Earth every 90 minutes at 17,000 mph with the first Nvidia H100 in space.
Co-Founder & Chief Engineer Adi Oltean walks through Starcloud 2 as the team fires the latches that deploy its 120 ft solar wingspan. It launches in January with 100x the power of Starcloud 1.
Co-Founder & CTO Ezra Feilden shows the 16-ton secondhand vibe table. The team bought it for $200K, instead of paying $1.5M & waiting 2 years for a new one, they rebuilt it over a weekend.
Engineering Manager Bailey Montaño leak checks the ion thruster after vibe. Bailey designed the radiators, which have 10x less mass & cost 500x less per watt than the ISS’s.
In the fab shop, Ezra shows the Starcloud 3 demonstrator, a 2-ton, 235 kW vehicle that fits 50 to a Starship.
Camilo Lopez cuts stainless on the laser cutter.
Engineering Operations Lead Miguel Rivera, formerly of Helion Energy, teaches me to MIG weld.
The tour ends in the new campus, about 100,000 sqft built to produce 100 satellites a week toward an 88,000-satellite constellation.
𝐓𝐈𝐌𝐄𝐒𝐓𝐀𝐌𝐏𝐒
(00:00) Philip Johnston, Co-Founder & CEO of Starcloud
(00:53) Tracking Starcloud 1 live in orbit
(02:46) Fixing a 16-ton vibration machine in one weekend
(06:59) From founding to orbit in 21 months
(08:17) From SpaceX to Microsoft to Starcloud
(09:49) Inside Starcloud 2
(13:12) A $15M mission to the nearest star
(17:49) What goes on a spacecraft to Alpha Centauri
(19:10) Why SpaceX could be worth $100 trillion
(20:41) “This isn’t a hot take, this is just a fact
(24:22) How big is Starcloud 3?
(26:41) How do you book a ride to space?
(30:11) How Starcloud tests in zero gravity
(33:49) Can GPUs survive space radiation?
(35:46) The first AI model trained in space
(38:40) Will solar satellites block the sun?
(40:57) How Starcloud tests a satellite for launch
(46:45) The paperwork behind a launch
(53:07) Molly puts her welding skills to the test
(1:00:18) The engineer behind Starcloud's radiators
(1:03:07) Inside Starcloud’s new campus
(1:05:55) Why launch slots are the new scarce resource
(1:06:45) Can someone shoot down a data center in space?
$TSLA — Philip Johnston says SpaceX could exceed $100T in 10 years and will buy Tesla.
“Elon posted that they're going to be worth $100 trillion in 10 years.
I think they might be worth more than $100 trillion in 10 years.
If all goes to plan and there's no world war, you believe SpaceX will be worth more than $100 trillion?
Correct.
What would make that up though? Break that down. How do you think about it?
Actually, well, mostly a good chunk of it will come from the fact that…”
@sourceryy @mollysoshea
Watch the full pod:
BREAKING: Inside Starcloud, the $2.3B Startup Putting Data Centers in Space
FACTORY TOUR - Redmond, WA
Primes quoted them $75M-$100M to build their first data center in space. They did it for $2.5M.
In just 21 months, Starcloud is one of the fastest companies to go from founding to orbit.
With $450M in total funding from Manhattan West, Nvidia, Cisco, Benchmark & EQT, the team walks me through their mission for AI in space.
I toured the new HQ w/ CEO @PhilipJohnston + co-founders @EzraFeilden & @AdiOltean
AI compute on Earth is limited by power & permitting. @Starcloud_ is moving it to orbit, where solar power is nearly constant. Philip says the dawn-dusk orbit alone could fit about 10 TW of compute, roughly 20x the entire US power grid. Once launch costs fall below $500/kg, he says compute in space will be cheaper than on the ground.
From Starcloud 1, Starcloud 2, to Starcloud 3, we saw:
Tracking Starcloud 1 live in orbit, circling Earth every 90 minutes at 17,000 mph with the first Nvidia H100 in space.
Co-Founder & Chief Engineer Adi Oltean walks through Starcloud 2 as the team fires the latches that deploy its 120 ft solar wingspan. It launches in January with 100x the power of Starcloud 1.
Co-Founder & CTO Ezra Feilden shows the 16-ton secondhand vibe table. The team bought it for $200K, instead of paying $1.5M & waiting 2 years for a new one, they rebuilt it over a weekend.
Engineering Manager Bailey Montaño leak checks the ion thruster after vibe. Bailey designed the radiators, which have 10x less mass & cost 500x less per watt than the ISS’s.
In the fab shop, Ezra shows the Starcloud 3 demonstrator, a 2-ton, 235 kW vehicle that fits 50 to a Starship.
Camilo Lopez cuts stainless on the laser cutter.
Engineering Operations Lead Miguel Rivera, formerly of Helion Energy, teaches me to MIG weld.
The tour ends in the new campus, about 100,000 sqft built to produce 100 satellites a week toward an 88,000-satellite constellation.
𝐓𝐈𝐌𝐄𝐒𝐓𝐀𝐌𝐏𝐒
(00:00) Philip Johnston, Co-Founder & CEO of Starcloud
(00:53) Tracking Starcloud 1 live in orbit
(02:46) Fixing a 16-ton vibration machine in one weekend
(06:59) From founding to orbit in 21 months
(08:17) From SpaceX to Microsoft to Starcloud
(09:49) Inside Starcloud 2
(13:12) A $15M mission to the nearest star
(17:49) What goes on a spacecraft to Alpha Centauri
(19:10) Why SpaceX could be worth $100 trillion
(20:41) “This isn’t a hot take, this is just a fact
(24:22) How big is Starcloud 3?
(26:41) How do you book a ride to space?
(30:11) How Starcloud tests in zero gravity
(33:49) Can GPUs survive space radiation?
(35:46) The first AI model trained in space
(38:40) Will solar satellites block the sun?
(40:57) How Starcloud tests a satellite for launch
(46:45) The paperwork behind a launch
(53:07) Molly puts her welding skills to the test
(1:00:18) The engineer behind Starcloud's radiators
(1:03:07) Inside Starcloud’s new campus
(1:05:55) Why launch slots are the new scarce resource
(1:06:45) Can someone shoot down a data center in space?
$MSFT — .@chamath says AI firms' debt needs rival 30% of Treasury issuance through 2027.
“So the AI complex now owes 340 to 400 billion dollars to the market. And you'd say, okay, why is that important? The problem is this. These companies are raising so much money, it starts to crowd out the capital markets. And the most important part of the capital market that it starts to crowd out is the appetite for government treasuries.
And what you can see here is that when you look at these…”
@theallinpod
Watch the full pod:
POD UP! 🚨
-- Anthropic S-1 Shocker: Greatest Business Ever?
-- AI's Biggest Financial Risk
-- Google’s $500B Wipeout
-- Midterm Predictions
(0:00) Bestie Intros
(3:22) Anthropic S-1 Breakdown: Greatest Business Ever?
(32:47) AI's Biggest Problem and Anthropic's Doomer Showdown
(1:01:57) Google Sheds $500B Market Cap in Worst Day Ever: What Went Wrong?
(1:14:17) Final Midterm Predictions and Scenarios
$MSFT — .@chamath says AI firms' debt needs rival 30% of Treasury issuance through 2027.
“So the AI complex now owes 340 to 400 billion dollars to the market. And you'd say, okay, why is that important? The problem is this. These companies are raising so much money, it starts to crowd out the capital markets. And the most important part of the capital market that it starts to crowd out is the appetite for government treasuries.
And what you can see here is that when you look at these…”
@theallinpod
Watch the full pod:
POD UP! 🚨
-- Anthropic S-1 Shocker: Greatest Business Ever?
-- AI's Biggest Financial Risk
-- Google’s $500B Wipeout
-- Midterm Predictions
(0:00) Bestie Intros
(3:22) Anthropic S-1 Breakdown: Greatest Business Ever?
(32:47) AI's Biggest Problem and Anthropic's Doomer Showdown
(1:01:57) Google Sheds $500B Market Cap in Worst Day Ever: What Went Wrong?
(1:14:17) Final Midterm Predictions and Scenarios
$ANT — .@DavidSacks says Sacks says Anthropic's 80B loss is non-cash; IPO could rival SpaceX's.
“It could very well be. I mean, I think this is probably going to be the biggest venture outcome of all time. Especially given the time in which this was created. I mean, this is only a six-year-old company. I think the exit here is going to be on par with SpaceX's IPO, but SpaceX has been around for like 20 something years. So this is probably the fastest value creation of any venture-backed…”
@theallinpod
Watch the full pod:
POD UP! 🚨
-- Anthropic S-1 Shocker: Greatest Business Ever?
-- AI's Biggest Financial Risk
-- Google’s $500B Wipeout
-- Midterm Predictions
(0:00) Bestie Intros
(3:22) Anthropic S-1 Breakdown: Greatest Business Ever?
(32:47) AI's Biggest Problem and Anthropic's Doomer Showdown
(1:01:57) Google Sheds $500B Market Cap in Worst Day Ever: What Went Wrong?
(1:14:17) Final Midterm Predictions and Scenarios
Intel ($INTC) sank about 29 percent today, its worst day since 1974, after weak guidance tied to a slump in PC demand.
• Reddit fell about 29 percent as Google's AI search cut its traffic
• Nasdaq closed down 2.9 percent, its worst month since March 2025
• Nvidia, Broadcom and AMD slid as AI spending doubts resurfaced
Key takeaway: AI spending payoff is now the market's main question.
An advance of 79% has materialized for $CDNA since Aug 24, 2026, accompanied by a 19.0% single-day gain.
The initial mention on X came from @BlogJulianKomar on Aug 24, 2026, when the price was $47.50.
Worth a follow: @BlogJulianKomar.
Their full record: fintwit.ai/analyst?id=114498…
Most accurate analyst accounts on X this week, by buy calls we track (scored on price since the call):
@RonShamgar 83% (5/6 calls up)
@Brownmoose 70% (7/10 calls up)
@alc2022 70% (7/10 calls up)
@7StarMike 69% (9/13 calls up)
@FIREDUpWealth 61% (19/31 calls up)
All five are worth a follow.
Every call scored: fintwit.ai/analyst
$MSFT — Martin Casado says Legal software is a shadow labor market far bigger than its spend.
“is your time tracker, right? Because you're billing by the hour. If you're in contingency law, it's like, well, your time matters, but you're not billing your customers on that. You probably need to have each case, here's how much money I'm spending on the case. But I imagine your primary tool is Microsoft Word. And this is how VCs would have gotten this industry wrong before, which is like, oh,…”
@a16z
Watch the full pod:
If a case won't earn a lawyer at least $5,000 in fees, most won't take it. AI is starting to change that math.
Eve co-founder and CEO Jay Madheswaran joins a16z's Alex Rampell to explain how AI can automate plaintiff law from the first client call through settlement, and how one firm used Eve to go from about 100 clients per attorney to 500.
They also discuss why legal software is secretly a labor market, why a single case can take 10,000 back-and-forths, why agents may soon negotiate with other agents, and what happens to the legal system when filing a case gets dramatically cheaper.
0:51 Plaintiff law vs. Big Law
3:50 Why lawyers won't take a small case
8:42 How AI grows a law firm's revenue
13:57 From 100 to 500 clients per attorney
15:53 Legal software's hidden labor market
20:25 Why Eve beats raw AI models
27:00 When both sides have AI
29:13 What's left for human lawyers
33:01 A lawsuit filed by AI in 4 hours
35:15 Can a small firm hit $1B in revenue?
39:16 Law's shift to outcome pricing
46:08 10,000 touches in one case
49:37 How AI beat the Thanksgiving trick
YouTube: youtu.be/iRMmPjunm3Q
@jayanthmadhesw1 @arampell
Alex Rampell says AI forces SaaS from per-seat pricing toward paying for outcomes.
“Well, the funny thing is that there's this movement underway to outcome-oriented pricing.
So like software, SaaS used to be per seat per month. And I used to joke, it was like the tall, grande event team model of software selling. It's like you go to like any SaaS company's website. This is why Claude can clone them all so readily. They all look the same. It's like $79 a month, $99 a month, $129…”
@a16z
Watch the full pod:
If a case won't earn a lawyer at least $5,000 in fees, most won't take it. AI is starting to change that math.
Eve co-founder and CEO Jay Madheswaran joins a16z's Alex Rampell to explain how AI can automate plaintiff law from the first client call through settlement, and how one firm used Eve to go from about 100 clients per attorney to 500.
They also discuss why legal software is secretly a labor market, why a single case can take 10,000 back-and-forths, why agents may soon negotiate with other agents, and what happens to the legal system when filing a case gets dramatically cheaper.
0:51 Plaintiff law vs. Big Law
3:50 Why lawyers won't take a small case
8:42 How AI grows a law firm's revenue
13:57 From 100 to 500 clients per attorney
15:53 Legal software's hidden labor market
20:25 Why Eve beats raw AI models
27:00 When both sides have AI
29:13 What's left for human lawyers
33:01 A lawsuit filed by AI in 4 hours
35:15 Can a small firm hit $1B in revenue?
39:16 Law's shift to outcome pricing
46:08 10,000 touches in one case
49:37 How AI beat the Thanksgiving trick
YouTube: youtu.be/iRMmPjunm3Q
@jayanthmadhesw1 @arampell
Bing Gordon says Bing Gordon rejects AI doom, predicting positive outcomes despite transition problems.
“I'm not a Doomer. I think Doomers are going to doom. If you wake up in the morning and you're cranky, AI is going to kill everything. If you're a journalist and you notice you get three times more clicks for bad news and good news, AI is going to kill everything. In history, technology, even nuclear fusion, no matter how scary it is, it has always uplifted people sometimes with a little bit of a…”
@MoreorLessPod
Watch the full pod:
@bingfish was there at the founding of Electronic Arts in 1982, spent 18 years at Kleiner Perkins, and sat on Amazon's board for 15.
He was also on Zynga's board, which is why it lands when he says players felt relief when they quit Farmville while retention and revenue were still climbing.
So where's the line between good game design and gaming your users? His answer takes in TikTok doom scrolling, WhatsApp as the rare product nobody complains about overusing, the sponsored products Amazon still shows Prime members, and ChatGPT ending every answer with one more question.
Then @davemorin asks what the agent era keeps circling: what gets worse if your AI wins for you?
"Friction is what makes pearls. If you don't get a sand in the oyster, you just get an oyster." And the next frictions, he says, come from edge conditions nobody tested for.
Plus the wow/WTF scale he used to judge EA's Hall of Fame games, and the bet-your-job letter he kept in a drawer.
@brit hosts. New episode out now.
Topics:
01:02 No Lessins this week: Brit and Dave bring in Bing Gordon
02:42 Amazon's board, Prime, and 18 years at Kleiner Perkins
05:03 Disneyland inside the computer
06:51 The classroom question that led to Electronic Arts
09:44 Raising $2M in 1982, then Atari collapsed
10:56 Vision, and why people think you're stupid
11:37 Bathroom OKRs and why goals beat perfection
13:13 Word of mouth, and why coin-operated hires leave
14:44 "Can a computer make you cry?" and hiring on mission
16:53 Why games retain: where everybody knows your name
17:34 Which social network did it best?
18:36 LinkedIn endorsements and getting people to come back
20:19 Farmville: players felt relief when they quit
21:18 Growth hacking is human growth hormone
23:04 Amazon, TikTok, ChatGPT: who's gaming their users today?
24:21 What gets worse if your AI agent wins for you
25:03 Friction is what makes pearls
26:30 Hugging Face, Cruise, and the edge conditions nobody tests
29:04 Writing Everybody Wins with his daughter Chloe
32:19 The visionary and the operator
33:20 The wow/WTF scale, and how to manufacture a wow
36:08 Wow moments in games, Facebook, Instagram and Amazon
38:42 Creating wow moments as a parent
41:28 The bet your job moment
42:23 You only got fired for blaming someone else
43:19 Good risk vs bad risk
43:53 Invest and express: Twitch, chat, and kids' art on the wall
45:59 Dave's favorite Bing-ism: the airplane aisle test
47:30 The best retention hack ever made is a wedding ring
48:01 Session economics, novelty, and leveling up
51:51 How to gamify a board meeting
54:03 Why a founder who loses the board twice gets replaced
55:23 Family board meetings and the vice president of fun
56:17 Lightning round: books, heroes, best investment
58:41 Bing's p(doom): doomers are gonna doom
59:36 The three things that make a good day
1:01:59 Dave's iPhone Duo verdic