Mach33 Financial Group

We research and invest in companies developing expansion technologies shaping the future of humanity, including space, AI, robotics, and advanced infrastructure

[Mach33 Financial Group @mach33]

Launch Keystone is out and you can find it here.
https://t.co/LrAhnZTsDW

[Aaron Burnett @aaronburnett]

What will it cost to launch a kilogram into space in 2040, if you don't own the rocket? We took a stab at forecasting customer launch price over the next decade.

We try very hard to account for everything in a bottom up model. Built off 77k rows of cleaned historical data, more than 500 assumptions and then monte carlo'd 1,999 times. hopefully, we are less wrong then the next guy, but we'll see.

Core takeaways:

Starship opening to customers and when it does is 'the hinge' in the model. no one else has a viable path to the scale of reusable tonnage to orbit to effect customer prices.

Internal usage (like Starlink/Starmind for SpaceX, or similar constellations for others like Terrawave/Blue Origin) dramatically dominates the total upmass to orbit driven by reduced internal costs.

Even with the dropping customer prices the difference in $/kg between internal and external uses is dramatic. due to things like payload optimization.

Interestingly enough, this model produces a fairly negative case (for customers purchasing launch) in the late 2030s where starship capacity is driven away from customers and toward Moon/Mars and internal endeavors causing a demand shock.

You can read about the model, core assumptions and it's build methodology here...
https://t.co/VelGNT8ciE

[Mach33 Financial Group @mach33]

use the mcp, data set on us.

[Aaron Burnett @aaronburnett]

these 3 animated charts we are completely built with Grok @bot leveraging the @mach33 MCP.

I asked grok bot to pull our global launch and payload manifest and make compelling animated charts on brand for Mach33. the process is exceedingly simple and addictive.

I fully expect if you try you'll find yourself spending more time and tokens then you expected (in a good way)

We've open sourced the underlying historical data (for a limited time) that i used for these charts. you can access that data set here and make your own.
https://t.co/XVtnjx9vTj

over 7,000 rows of launches and 27,000 rows of payloads. nearly limitless ways to cut and display the data. creativity is really the only limiter.

[Aaron Burnett @aaronburnett]

26 v3 Sats, Payload Optimization, Flight 14 | Episode #20 https://t.co/sQDYgMy4Y2

[Mach33 Financial Group @mach33]

26 v3 Sats, Payload Optimization, Flight 14 | Episode #20 https://t.co/Uc8y6E7TAF

[Aaron Burnett @aaronburnett]

when spacex leadership says starship is the priority, this is what they mean. Our forecasts through the 2030s (conservative compared to company targets), suggests that SpaceX will put up more than 4.2M tonnes into space by 2040.

These charts below compare SpaceX's internal projected tonnage vs external (customer paid for launch) by 2040.

the lines on this chart look much closer than they are because its in log scale, on purpose so you can actually see some of the inflection points. (read the end state numbers if you want a sense of how much we forecast internal to outweigh external)

The 2,000 run monte carlo placed the total upmass range between 1.9M t (p5) and 8.3M t (p95).

What it shows is exactly how important starship is to hoisting modularized cell towers (starlink) and modularized super compute racks (starmind) into space.

And how important Starship opening to commercial launch buyers is to the external mass. This has massive down stream effects on the industry.

these are charts are built off the Mach33 Launch Keystone used by institutions for understanding and predicting the next decade of access to space.

You can read all about the Keystone here.
https://t.co/VelGNT8ciE

[Aaron Burnett @aaronburnett]

Launch Keystone, Stoke Raise, empire building? | Episode #19 https://t.co/Inzmx0nJ4C

[Mach33 Financial Group @mach33]

Launch Keystone, Stoke Raise, empire building? | Episode #19 https://t.co/4M2inaa3fq

[Mach33 Financial Group @mach33]

live in 5

[Aaron Burnett @aaronburnett]

Going live in 5 mins talking our predictions on launch for the next decade. and why our numbers point more to empire building than railroads.

you can catch it on my feed or @mach33 feed.

[Aaron Burnett @aaronburnett]

we looked at 70 years of launch and compiled some of our most interesting learnings. this weeks brief pulls all that depth into an ultra scannable and shareable mini-site entirely open to the public

https://t.co/i33fqdFjpW

A couple deep cut insights that i found most interesting.

1) [video] 25 SpaceX boosters flew more times than every other countries and entity in the world combined. flying a total of 162 times compared to the rest of the world flying 157 times.

2) [chart] V2 Mini Optimized is so mass-optimized that 2026 Starlink tonnage is on pace down 7.3% while nameplate bandwidth is up 10.2%. this doesn't account for V3. just showcases the power of payload optimization.

the only way you can hope to accurately predict the future is if you intimately understand where we have been.

[Aaron Burnett @aaronburnett]

Orbital Failure Rates creates interesting context for global historical launch. Really puts into perspective how difficult reaching space is, and how impressive SpaceX has been in quickly achieving the lowest cumulative orbital failure rate in history.

in case you think i'm selecting datasets to paint a more rosey picture for Spacex. i've included the full table of entities and their failure rates in the comments.

[Aaron Burnett @aaronburnett]

Visualizing Starlink's bandwidith growth by satellite generation helps you see how seemingly incremental improvements can lead to significant inflections. SpaceX Starship will make this curve appear flat soon.

[Mach33 Financial Group @mach33]

How do defense cos share their wins when their capabilities are potentially a matter of national security?

[Aaron Burnett @aaronburnett]

There is a very sharp elbow coming to Starlink's bandwidth charts, but the vast majority of investors are treating it as incremental or simply ignoring it.

I find it helpful to use recent history to illustrate a point. Luckily we have data to show the last Starlink Inflection, and I believe its highly likely to closely illustrate what's coming for us in the next ~12 months.

[first screenshot] you can see that in Feb 2023 Starlink transitioned from V1 to V2. At that time their global bandwidth was ~64 Tbps. Though linear, the historic starlink bandwidth growth looked quite impressive from 2020-2023.

The V2 mini sat was a substantial bandwidth increase ~5.5X more bandwidth per sat, but only ~2.1X bandwidth per launch because the sats we're more massive.

[second screenshot] shows one year of V2 deployments. 187 Tbps was a 3X of global starlink bandwidth in one year. Due to V2's ~2.1X bandwidth per launch and an near doubling in launch cadence.

[Video] you can see this trend continue and accelerate with V2 Optimized from 2024 to 2026, to the point that it makes the first 3-4 years of starlink bandwidth look flat, the same growth that looked sharp in 2023.

Bandwidth growth matters because it's the supply side of the market dynamic.

The historical relationship between Starlink bandwidth growth and customer growth has been correlated even if not completely causal.

2022 -> 2026
'22 Customers 1M | Bandwidth 57 Tbps
'26 Customers 10M | Bandwidth ~600 Tbps

Clearly both customers and bandwidth grew an order of magnitude over that period. To be clear there has been corresponding decrease in consumer's price for connectivity that plays a role making the customer growth number not pure 1:1 to bandwidth growth.

BTW - this was the point of Elon's earning statement about 100X bandwidth at 1/10th the cost to consumer/enterprise still driving toward ~$200B in revenue. there likely won't be 90% cost reductions inside 5 years, but if they did somehow that would be very good for consumers and enterprises.

Fast forward to this month and the main point I was trying to make at the beginning. Starlink V3 is clearly another inflection, if underappreciated.

We are looking at potentially the first operational deployment of V3 starlink sats in September.

how does a the V2 -> V3 jump compare to the V1 -> V2 jump?

V2 -> V3 bandwidth/sat:
V2 is 96 Gbps
V3 is ~1,000 Gbps

the V2:V3 jump (10X) is twice the size of the V1:V2 jump (5.5x) on a per sat basis. What about the starship of it all?

V2 -> V3 Sats/Launch:
V2 had ~28 on a F9
V3 has 20-60 on a Starship

The V2:V3 jump compounded by the F9:Starship means the bandwidth per launch is 7x-22x, compared to 2.1X per launch for V1:V2 jump.

another way to say this is that V2:V3 bandwidth acceleration over the next 12 months will be 3-10X greater than the V1:V2 inflection from Feb 2023-Feb 2024.

Once you appreciate this mathematical certainty the only questions that remain is how much do prices need to drop (if at all) to keep customer growth and bandwidth growth as close to 1:1 as possible.

History rhymes and this history lesson points to the fact that Starlink is a generational cash engine and will fund and functionally connect the AI compute build out.

I'd suggest watching the video to get a sense of how inflections play out in realish time.

[Aaron Burnett @aaronburnett]

CyberCab getting an integrated Starlink and all the corresponding capability upgrades wasn't a viable option even 2-3 years ago. That's how fast Starlink bandwidth supply has grown.

for context, 7 years ago there were only 66 broadband satellites in low earth orbit from anyone. Since that point total broadband satellites have exploded to more than 12,500 satellites.

What's clear from this race is what even just a 3-5 year launch advantage translates to in usable capability like bandwidth.

also notice how much iteration Starlink has done with its variants, which has translated to dramatic bandwidth improvement. in a couple years the Starlink V3s will quickly dominate (and will need to do a bandwidth version of this chart as well)

[Aaron Burnett @aaronburnett]

Sputnik to SpaceX Launch Manifest, 3.5T Revenue | Episode #18 https://t.co/Yi6GmF880W

[Mach33 Financial Group @mach33]

Sputnik to SpaceX Launch Manifest, 3.5T Revenue | Episode #18 https://t.co/02LiTKSsXF

[Aaron Burnett @aaronburnett]

Since 1957 more than 22,196 tonnes of usable payload has been deployed to space by more than 20 countries and private companies.

For 69 of the last 70 years, Russia (including the Soviet Union) led the world in cumulative usable payload.

According to our global launch manifest, that changed in 2026 when SpaceX tallied 7,962 T of usable payload into orbit as of July 2026, driven heavily by Starlink deployments.

Based on current trajectory, it seems unlikely SpaceX will be giving up their lead anytime soon.

The Mach33 global launch manifest has 7,117 rows of launch and 27,537 rows of payloads. Our premium analysis this week looked at all 70 years of launch.
https://t.co/YLTMgZBDo2

This animation compares usable payload mass, which removes the Shuttle Orbiter Mass to correct for actual usable mass deployed.

Grok Bot was used to animate the Mach33 data. Also Grok Bot was used to compose an original symphony for the music on this animation.

[Aaron Burnett @aaronburnett]

moving our lives to X has increased our viewership an order of magnitude and growing every week.

honored that anyone cares enough to listen to us geek out on space and SpaceX research.

One thing we need (that i know X is working on) is better live comments section. The only thing I miss about Zoom recordings was more engaged questions from the watchers.

[Mach33 Financial Group @mach33]

we are live!

[Aaron Burnett @aaronburnett]

Starbase Louisiana, Florida Starship, more Grok Bot, Propellant Math | Episode #17 https://t.co/4tE8i2jKm0

[Mach33 Financial Group @mach33]

Starbase Louisiana, Florida Starship, more Grok Bot, Propellant Math | Episode #17 https://t.co/ApVlD5czw4