In memory of Mate Marušić (@paracho1) header
In memory of Mate Marušić (@paracho1) profile picture

Veni Vidi Vici ! Finance, Financial Markets, Art, World History, Languages, Classical Music . $TSLA investor and fan.

Toronto, Ontario, Canada
Joined October 2010
Photo by In memory of Mate Marušić (@paracho1)Photo by In memory of Mate Marušić (@paracho1)Photo by In memory of Mate Marušić (@paracho1)Photo by In memory of Mate Marušić (@paracho1)Photo by In memory of Mate Marušić (@paracho1)Photo by In memory of Mate Marušić (@paracho1)Photo by In memory of Mate Marušić (@paracho1)Photo by In memory of Mate Marušić (@paracho1)Photo by In memory of Mate Marušić (@paracho1)
In memory of Mate Marušić retweeted
Elon gave two talks today at DealBook and TomorrowX Summit, revealing more about SpaceX's plans than most earnings calls. Here are the 11 key takeaways: 1. SpaceX is scaling AI compute at an extraordinary pace: SpaceX is adding roughly 500 MW monthly, equivalent to 220,000 GB300s. Elon expects over 1 GW monthly by end-2027: a 6 GW annualized pace today, doubling to 12 GW. 2. Its build costs are far below industry estimates: Elon says outside estimates are off by up to 2x, with SpaceX building at roughly 60% of industry costs. Its in-house "Army X" builds the data centers, while Tesla Megapacks smooth training power. I'd already been modeling a 25–30% SpaceX cost advantage. 3. Anthropic could become a 10 GW customer: Anthropic told Elon its most likely outcome is renting another 10 GW from SpaceX, on top of existing capacity. Contracts allow termination with 3–6 months' notice, but Anthropic says staying is "extremely likely." This reinforces my model that Anthropic will take most upcoming capacity. 4. SpaceX is targeting a vastly larger telecom market: Elon argues AT&T and Verizon won't disappear as bandwidth demand explodes. Humans communicate at roughly 100 bits per second, while superintelligence and robots could require a trillion. SpaceX is targeting a market orders of magnitude larger than existing telecom revenue, as I'd modeled. Even more interestingly, this creates an opportunity to push its AI business further. 5. Its next-generation satellite network arrives next year: Direct to Cell V2 launches with 40x V1's bandwidth. Micro cell towers using 800 MHz and rooftop units resembling Starlink dishes will extend indoor coverage, with rooftop hardware entering production next year. Starlink is becoming a full scale carrier. 6. SpaceXSI has rebuilt its entire AI software stack: Over six to eight months, it rewrote training, reinforcement learning, and inference from scratch, close to the metal in C and C++. Elon thinks it "might have the best training stack software in the world." Grok 4.8 is also being built for practical usefulness rather than benchmark leadership, which is great for Grokbot. 7. Terafab could become the largest project on Earth: The facility is planned to reach 100 million square feet, with phase-one volume production targeted for late 2028. Elon says established chipmakers are risk-averse because of massive capex, while SpaceX will take risks and iterate rapidly toward a better chip process by 2028. 8. Optimus is moving toward industrial work: Elon expects an unstaged demonstration in a month or two and says Optimus could do dishes within 12–18 months. Its first major industrial role will be mining and refining ore for national security. 9. A Tesla–SpaceX merger is becoming increasingly logical: Elon says the companies are "probably" better together as AI and robotics bring them closer. Both have AI software teams and advanced chip-design capabilities, while superintelligence compute could eventually move into orbit. Shareholders will decide; Elon says he shouldn't receive an award simply for merging them. 10. Ship 42 could accelerate the entire Starship reuse program: Ship 42 needed only minor repairs before flying again, but SpaceX is retiring it for public display. This brings rapid reuse much closer. 11. SpaceX is targeting 200 GW of annual solar manufacturing for space: That's roughly 40% of average US electricity consumption. Its Starmind compute unit delivers the same capability at roughly a quarter the size of a Vera Rubin NVL72 rack, which is a win in space. Elon says it's also simpler than Starlink V3, which is already in production. Simply put, SpaceX is building toward a trillion-dollar annual revenue business within just a few years, at scale never seen before. Many of these numbers I've already modeled and shared over the past few months. Yesterday's announcements bring those projections closer to reality. The scale of what's coming is extraordinary!
112
585
61
3,984
432,551
In memory of Mate Marušić retweeted
“An intellectual man in complete solitude has excellent entertainment in his own thoughts and fancies.” Arthur Schopenhauer
21
160
13
1,221
23,282
In memory of Mate Marušić retweeted
BREAKING: SpaceX adds ~$23,000,000,000.00 as Anthropic goes public Elon Musk says Anthropic wants 10 more gigawatts of SpaceX computing power
28
78
8
1,152
62,757
In memory of Mate Marušić retweeted
"Change is never painful, only the resistance to change is painful" * Gautama Buddha
7
43
3
209
2,942
In memory of Mate Marušić retweeted
“I am not what happened to me, I am what I choose to become.” ~Carl Jung
43
428
46
2,147
44,518
In memory of Mate Marušić retweeted
Greed can be satisfied. Envy, cannot. @naval
6
43
3
533
19,341
In memory of Mate Marušić retweeted
JUST IN: Anthropic researcher predicts AI will leave nearly every human "permanently unemployable" & reliant on government welfare to survive.
1,037
1,324
744
12,575
1,354,261
In memory of Mate Marušić retweeted
BREAKING: Elon Musk reveals SpaceX will start launching satellites next year that connect straight to regular smartphones. 49% chance SpaceX becomes a US mobile carrier by end of next year. 🔗 polymarket.com/r/7mscHFk
8
82
7
879
18,507
In memory of Mate Marušić retweeted
JUST IN: Elon Musk says SpaceX and Tesla would "probably" be better together 25% chance they merge this year
39
41
6
1,188
42,615
In memory of Mate Marušić retweeted
𝗦𝗽𝗮𝗰𝗲𝗫 + 𝗧𝗲𝘀𝗹𝗮: 𝗮 𝗵𝘆𝗽𝗼𝘁𝗵𝗲𝘀𝗶𝘀, 𝗻𝗼𝘁 𝗮 𝗱𝗲𝗰𝗿𝗲𝗲 Elon on a merger: the two companies don’t need to be together. Would they be better together? Probably. But he isn’t treating that as a done deal. He’s testing it. Do they execute better and create more value combined than apart? Does that claim hold up under scrutiny? He has a duty to the other shareholders and can’t drag them into a deal that’s bad for them. A Tesla vote may or may not pass. So the shareholders decide. What has changed is the input: areas of convergence that didn’t exist before, and they matter more and more to the future.
5
7
2
66
21,851
In memory of Mate Marušić retweeted
2X LONG AND 2X SHORT ANTHROPIC ETFS LAUNCH MONDAY RIGHT AFTER THE IPO TRADR IS ROLLING THESE OUT AND HERE ARE THE TICKERS $ANTT - 2X LONG ANTHROPIC ETF $ANTZ - 2X SHORT ANTHROPIC ETF BE CAREFUL, LEVERAGE INVOLVES SIGNIFICANT RISK. READ MORE AT TRADRETFS.COM
26
34
7
283
178,062
In memory of Mate Marušić retweeted
JUST IN: Man uses Meta's own AI to file a $75,000.00 lawsuit against Facebook after being banned, sending over 100 emails, contacting the FBI, and faxing the White House
39
54
12
798
66,069
In memory of Mate Marušić retweeted
Zapovjednik Bojnog zrakoplovstva, pukovnik Adalbert Rogulja često se i sam uvjerava o stanju oružja, kojim se njegovi zrakoplovci bore. The Commander of the Air Force, Colonel Adalbert Rogulja himself checks on the condition of the weapons his airmen use to fight.
2
21
371
In memory of Mate Marušić retweeted
50
2,410
101
14,536
169,907
In memory of Mate Marušić retweeted
Is the Thucydides Trap ever avoidable? I wish it were, but, inevitably, the most interesting outcome is the most likely.
3,670
2,484
741
28,040
7,676,382
In memory of Mate Marušić retweeted
2
46
3
235
3,861
In memory of Mate Marušić retweeted
“Sometimes you have to stand alone, not to prove you're right, but to prove you're not afraid of being wrong.” - Marcus Aurelius
18
452
14
2,809
37,517
In memory of Mate Marušić retweeted
2008年,马斯克曾邀请芒格投资特斯拉,当时公司估值只有约2亿美元。 芒格没有投。 据李录回忆,当年几人共进午餐,马斯克试图说服芒格投资。 没想到,芒格反而当着众人的面,逐条分析特斯拉为什么会失败。 更有意思的是,马斯克竟然没有反驳。 他后来回忆:“这让我很难过,但我承认他说的都对。我们很可能失败,但无论如何,这件事值得一试。” 后来的故事,大家都知道了。 特斯拉不仅活了下来,还一度成为全球市值最高的汽车公司。 但这并不意味着芒格判断错了。 芒格看到了风险,选择不投;马斯克同样看到了风险,却决定赌上一切。 两位顶级头脑,面对同样的风险,做出了截然不同的选择。 真正的智慧,不是所有人都做出同样正确的选择,而是清楚自己在追求什么,又愿意为此承担什么。 #芒格 #马斯克 #特斯拉 #投资
220
392
33
4,013
331,482
In memory of Mate Marušić retweeted
Seneca on virtue
2
19
207
2,428