Michel (@micmat02) header
Michel (@micmat02) profile picture

My investment diary.

Joined January 2010
Photo by Michel (@micmat02)Photo by Michel (@micmat02)Photo by Michel (@micmat02)Photo by Michel (@micmat02)Photo by Michel (@micmat02)Photo by Michel (@micmat02)Photo by Michel (@micmat02)Photo by Michel (@micmat02)Photo by Michel (@micmat02)Photo by Michel (@micmat02)
Think I met @Peter_J_Beck for the first time in 2015. My friend Al Weston told me about this self-taught engineer in New Zealand who was building rockets and actually seemed to know what he was doing. I was skeptical. I flew down to Auckland and, sure enough, Peter had three rockets already built and was clearly a one-of-a-kind human and engineer. Nothing about hitting 100 rockets in record time surprises me now. If you want to read more about Peter, I have a book - amazon.com/gp/aw/d/006299887… If you have less attention span, I have a movie - hbomax.com/movie/55ee21e5-42…
Black rocket 🤝 Golden milestone. Electron is all dressed up for launch 100.
14
31
3
290
10,966
Michel retweeted
SpaceX is effectively the best neocloud right now... SpaceX insists on short term leases only, so as long as compute keeps getting scarcer and the compute shortage curve keeps sloping up and to the right, SpaceX can keep raising prices. On top of that, SpaceX has the best infrastructure buildout capability in the industry and its own compute demand, so it can absorb excess demand at any time. That is why it commands a premium of 2x, and as much as 3x, over competitors like IREN. ($60 billion per GW) To track Nvidia's revenue, SpaceX is now the company to watch most closely. Until SpaceX ASICs come out of Terafab in 2030, it will devour more Nvidia chips, and more ravenously, than anyone else in the world. Wouldn't SpaceX easily absorb the slowdown in Nvidia chip demand caused by Google and other hyperscalers developing their own ASICs?
85
130
28
1,915
159,901
In 2017 a wise man told me: “When you think it’s over, it’ll just be getting started.” I think we’ll reflect upon 2026 as such a period for blockchains.
107
194
35
2,034
134,809
Michel retweeted
We announced an agreement to acquire a nationwide low-band spectrum license portfolio that will pave the way for @Starlink to become a major mobile carrier in the US. With this new spectrum and our Gen2 constellation, Starlink Mobile can ensure Americans have access to high-speed mobile broadband no matter where they are → spacex.com/updates#starlink-…
3,144
8,271
2,282
66,691
12,543,459
Love the smell of fear in an early bull
Perfect storm today I guess. Oil up, yields up, stocks down, a bunch of fud, and traders being idiots as usual.
65
176
29
2,124
116,605
The Internet Needs Crypto to Survive AI: youtube.com/watch?v=TKVY4hVA… One of the most exciting, and underrated, developments in crypto atm revolves around this @Bankless conversation with @eastdakota of @Cloudflare. ~20% of all websites are routed through Cloudflare, and in this conversation Matthew tells us that the business model of the Web is under attack. To put it simply, agentic traffic is exploding, and agents don't click on ads, but ads underpin much of the "free" stuff we get online. The infrastructure is creaking under agent load. Matthew's assertion is that microtransactions, at a scale not previously seen, will be needed to pay for access to content, as the advertising model of the Web falls apart with agents. Crypto people have talked about microtransactions forever, but it was for people, which never worked, whereas we're now contemplating it for agents. He says Cloudflare handles ~500M requests per second, and that 1-10% of those requests may be suitable for this microtransaction model. A blockchain environment is the only payment network that could theoretically handle this scale, and it would need to hit 5-50M TPS. 5-50M TPS in prod dwarfs anything yet attempted in crypto, and is far beyond anything TradFi can handle. There are very few blockchain environments that could conceivably provide this throughput. Just as important as the scale, is the model and trust assumptions. While Cloudflare may roll this out as a solution for its customers first, what's implicit here is the potential for an entirely new business model for the Web. If Cloudflare walks this path, I hope that what they build has permissionless validation, allowing for anyone to audit the network. Cloudflare has been a torch bearer for the open Internet, and so it only follows that if they're to pioneer a new business model for the Web, that the architecture be an open network. Doing so will engender the most trust, goodwill, and adoption in the long run. Web 3.0 is dead, long live Web 3.0.
54
55
18
453
76,402
Don’t care Remain bullish Bears can suck it
Don't care Remain bullish Bears can suck it
12
25
16
646
28,814
Michel retweeted
Cybercab is absolutely magical. Tesla has actually done it. A domestically mass-produced affordable autonomous electric vehicle is available to the general public. A car with no steering wheel generating revenue. It’s sad to come back to a city without Robotaxi after being in Austin. Like stepping back in time. The event was awesome. I know it was tough to follow online, but we all knew the story going in. This was a launch, not an unveiling. It felt like the city transformed immediately. Cybercabs were everywhere, you couldn’t take a trip without seeing another one. Outside of the ride experience itself, the scale of this launch is what I want to make sure people understand. On previous trips, Robotaxi felt like a test network (an amazing one). Not anymore. Suddenly it’s real. It’s hard not to laugh when you ride past a Waymo in a Cybercab. Good luck… As I’ve said before, Robotaxi is way smoother than Waymo. It feels natural. Supernatural, really. This holds true for Cybercab. Waymo feels like a computer. And it looks like one too. The production Cybercab is beautiful. It really is special. The color, the lines, the lighting, the doors. It looks great. While also being the most efficient car in history? Insane. I think it fits perfectly in the space between familiarity and the future. It doesn’t stick out, but it catches your attention. After exiting our first ride, there was a large group of 10-20 people. They were all talking about the Cybercab as we walked by. These doors are like gasoline on a word of mouth wildfire. Our rides were flawless. Better than a human driver. No anxious moments. Cybercab is very comfortable, even at 6’2. Leg room is particularly notable. It would be tough to find a car that has this much leg room for two passengers at the same time. No pedals, center console, or multiple rows of seating creates a lot of space and a very open feel. Robotaxi is unquestionably one of those products that will be way cheaper and way better than what existed before. I like those. No one is going to want to roll the dice on some random person and their car when the alternative is a private, safe, comfortable, clean, consistent, quiet, personalized Tesla with tons of room and a massive screen. It’s a premium experience on every level, and we get it cheaper thanks to technology. No noise or air pollution, either. I don’t have much to critique. Little refinements that I know will happen. Pick up and drop off were satisfactory, over time will become excellent. Grok integration is already impressive, and it’s clear that we’ll eventually be able to just give directions like “pull over around the corner up there.” You can already have Grok update your destination in Robotaxi, it pushes the update to the Robotaxi app on your phone for approval. I leave the event with more confidence in Tesla’s ability to scale, but also renewed patience. I have a greater appreciation for the intentionality of limitations (service area, times, fleet size) rather than seeing them simply as barriers that have yet to be overcome. I now believe the objective is to build slowly, safely, to critical mass. A tail-risk event is inevitable. The absolute last thing we need is some bad luck on a tail-risk event at an early stage before enough data exists to prove its rarity. That data is accumulating quickly and will become undeniable soon enough. Huge congrats to Tesla and Elon for this incredible milestone. Decades of work culminating in a life-saving, sustainable, affordable, economically exceptional product. It’s all pretty surreal, a testament to vision and determination. I love this company!
561
1,534
228
17,181
831,761
Eventually investors will connect the dots to Bittensor bittensor:native
OpenRouter, poolside, Hugging Face Everything open source getting acquired
97
345
55
1,793
370,287
$FCX Best in class name looks ready to go here... Give me $80. Watchlist!
43
17
4
274
215,894
Why this is a big deal: this mRNA therapy instructs the body to produce the specific fingerprints of the person's tumor and trains the immune system to hunt and kill it. First time in a Phase 3 trial. One way cancer grows is by hiding from the immune system and turning off attacking T-cells. This mRNA therapy provides the tumor's genetic fingerprint so that the immune system can train T-cells to identify the cancer. Keytruda then removes the brakes that cancer had put on the T-cells enabling them to eliminate the cancer.
490
523
103
10,141
840,196
I think @bot is another “Claude Code” moment for AI. I would estimate my personal AI usage is up something like 100x. And for everyone who reached out about how to build a “podcast summarizer” it took me about 15 seconds in Grok Bot and is better than what I had before.
381
381
119
5,887
6,396,153
Michel retweeted
Introducing Grok Bot, now in early beta. Bots are AI teammates that do real work for you. They sign in to your tools, use them just like you do, and come back with finished work.
3,758
4,542
3,481
39,577
58,918,417
You guys see what's going on, right? For decades and decades, the US government has chased this idea of responsive space. Rockets that can go up on very short notice, ideally from lots of locations, and place satellites over areas of interest and importance. It's been an incredibly hard thing to pull off. SpaceX started chasing this mission in a way but then moved away from the Falcon 1 and focused on Mars. Rocket Lab has very much picked up the mantle. It's putting pads in more and more places. It has two pads in spots rather handy for things happening in or near Taiwan. This is stuff Lockheed and Boeing and others thought about but could not even begin to make happen. Rocket Lab is basically becoming a space prime the likes of which we have never seen before.
65
162
42
1,827
144,787
Michel retweeted
Replying to @heeraniparkash
This is the kind of discourse the market likes especially nowadays. And to be honest a bit of a fresh air compared to what we’re used to see. If I had to guess, I believe the market will respond positively in the next days/weeks to RKLB.
1
1
3
85
Here are all the highlights from @SpaceX's Q2 2026 earnings report: • Total revenue: $7.81B, up 92% YoY from $4.07B and up 66% QoQ from $4.69B. • Net loss: $541M, improving by $467M YoY from a $1.008B loss. • Loss from operations: $143M, improving from a $970M loss YoY and a $1.94B loss in Q1. • Adjusted EBITDA: $3.53B, up 191% YoY from $1.214B and more than tripling from $1.127B in Q1. • Total Capex: $18.3B, up from $10.1B in Q1 and $2.8B a year ago, largely driven by AI infrastructure investment. • Space revenue: $962M, up 29% YoY and 55% QoQ; launch services contributed $648M and launch & development $314M. • Space operating loss: $542M, versus a $369M loss YoY and $662M loss in Q1. • Space Adjusted EBITDA: Loss of $205M, versus a $93M loss YoY and $351M loss in Q1. • Space Capex: $1.174B, up from $946M YoY and $1.052B in Q1. • Connectivity revenue: $4.29B, up 66% YoY and 32% QoQ; Consumer revenue was $2.48B (+44% YoY) and Enterprise & Government revenue was $1.8B (+108% YoY). • Connectivity operating income: $1.65B, up 79% YoY from $923M and up from $1.18B in Q1. • Connectivity Adjusted EBITDA: $2.597B, up 64% YoY from $1.583B and up from $2.08B in Q1. • Connectivity Capex: $1.36B, up 21% YoY from $1.130B. • Starlink subscribers: 12M, doubling YoY from 6M and increasing by 1.7M during Q2; ARPU remained $66/month. • Starshield: Awarded more than $6B in multi-year U.S. government contracts, primarily from two major Space Force contracts. • AI revenue: $2.56B, up 247% YoY and 213% QoQ; AI solutions & infrastructure revenue was $2.19B and advertising revenue was $367M. • AI operating loss: $1.257B, improving 49% QoQ from a $2.46B loss and from a $1.524B loss YoY. • AI Adjusted EBITDA: Positive $1.14B, versus a $609M loss in Q1 and $276M loss a year ago. • AI Capex: $15.8B, more than doubling from $7.7B in Q1 and versus $749M a year ago. • Cloud Services Agreements: Signed multiple agreements totaling $14.1B in contracted sales, generating $1.6B of incremental AI infrastructure revenue during Q2. • Backlog: $47.5B at the end of Q2. • Cash, cash equivalents & marketable securities: Approximately $100B at quarter-end ($93.5B cash + $6.48B marketable securities). • Total assets: $192.7B, up from $92B at year-end 2025. • Total debt & finance leases: Approximately $39.3B, consisting of $2.52B current and $36.8B long-term. • Launches: SpaceX completed 38 launches in Q2 - 10 customer and 28 internal launches. • Launches YTD: Reached 78 launches through June 30. • Mass to orbit: Deployed 485 metric tons in Q2 and 1,041 metric tons YTD. • Starlink constellation: Reached ~10,200 satellites in orbit as of June 30. • Starlink coverage: Expanded to 167 countries as of June 30. • Starlink aviation: Signed a major agreement with American Airlines and activated service on additional airlines including Southwest, Virgin Atlantic, Iberia and Aer Lingus. • Starlink Mobile: Launched new international carrier partnerships with SoftBank, NTT Docomo and Spark NZ. • Starshield: Awarded more than $6B in multi-year U.S. government contracts, primarily through two major Space Force contracts. • Spectrum: Received FCC approval for the EchoStar license transfer, covering 65 MHz of U.S. spectrum and certain global Mobile Satellite Service spectrum licenses. • AI compute: Expanded nameplate compute capacity to 1.4 GW, up from 1.0 GW in Q1 and 0.4 GW a year earlier.
60
299
30
2,264
146,784
Today, Tesla officially opened the world's first public Megacharger in Bloomington, California, about an hour east of Los Angeles. • 6 charging stalls • Each stall capable of 1.2MW (1,200kW) peak charging speeds • Open to all Tesla Semi customers
113
491
49
5,414
243,522
$RKLB 🚨UPDATE on the @Peter_J_Beck sale from the Form 4 filed today. @RocketLab’s CEO who cut his own salary to $1, waived every bonus, and voluntarily cancelled 392,155 unvested RSUs; handing back approx $30M in compensation so the capital could go to R&D instead. 👀 ‼️I’m seeing a lot of misinformation spreading that’s worth correcting. Many of it produced by quick Ai regurgitations without full context. Do your own research instead of trusting lazy posts‼️ First, the sale itself: SPB did NOT sell the full 5M shares. • 3.28M shares sold of 5M authorized • 1.72M left unsold when the window closed TODAY, 07/08/26 • Sold automatically over 3 days (July 6–8) under the pre-set plan disclosed in March • Prices ran from around $101 down to $81 as the plan sold into the pullback There was no “dump at the top”. ⚠️Second, the "he only has 2.2M shares left" claims: That's a MISREAD. A Form 4 only shows COMMON stock. The 2.2M figure (1.72M in the trust + 492K personal) is just the COMMON portion. The vast majority of SPB’s stake, roughly 40M+ shares, is held as CONVERTIBLE PREFERRED STOCK. It doesn't appear on the Form 4 table. It's disclosed in his Schedule 13D filed in March: total position of 46.4M shares, 7.51% of the company. The actual math: 46.4M held-3.28M sold=43.2M remaining SPB sold approx 7% of his stake and retains 93% of his position in the company he founded. #RocketLab #PeterBeck
$RKLB Growth sector downturn today with Space names taking an average 7% haircut again. A little extra for @RocketLab as folks digest the sale of shares by CEO Peter Beck. I am happy to add shares again at this level. That’s how it works. Some context on the Peter Beck share sale that many have already pointed out: This was NOT a surprise. It was disclosed back in March 2026 in SEC filings; a pre-scheduled Rule 10b5-1 sale set up months in advance, with a window closing July 8. TOMORROW. The details we've known since March: • Up to 5M shares, approx 10% of his holdings • Stated purposes: diversification, estate planning, philanthropy • Beck still retains approx 90% of his stake 👀 👀 Worth noting: the same week he filed the plan, Beck also cut his salary to $1, waived all bonuses, and voluntarily cancelled 392K unvested RSUs; redirecting that capital to company priorities and R&D. ‼️ This is his 2nd out of 2 major planned sale events so far (2025, mid 2026), all pre-scheduled under 10b5-1 plans; with one block trade sale in Sept 2023 which was his first open market sale ever. He’s also had 2 small automatic tax sales as sell-to-cover for vesting RSUs in November of 2025 and March of 2026. All this to say, that the man who is now earning ONE DOLLAR has sold a couple of other times and the sky has NOT FALLEN. The company has continued to execute and go up in value, and is more derisked than ever IMO. Ignore the noise. Inform yourself and make a decision on what to do during pullbacks. I’m doing what I’ve always done. NFA.
24
38
4
304
38,139
A few years ago João and I went on the Lex Fridman podcast and somehow ended up talking about one of the greatest sports rivalries ever: Messi vs Ronaldo. Today YouTube randomly recommended me the clip again. The algorithm knows! 🇵🇹
5
7
126
12,020
Michel retweeted
Walking the Optimus production line in Fremont
5,153
5,561
767
57,782
6,776,970