Starlink V3 satellites have >10X bandwidth of V2 and there’ll be >10X launched, which means >100X more bandwidth.
Also, altitude will be 350km vs 550km, so min latency can be cut in half.
Light travels 300km/ms in space, so physics round trip min latency drops to <5ms.
Ben söylemiyorum, dünyanın en değerli şirketlerinden birinin kurucusu söylüyor.
Anthropic kurucusu Dario Amodei’ye soruluyor:
“Eğer elinde bugün 100 dolar olsa ve kazanmak için tek bir hisse senedine yatırım yapmak zorunda kalsan, hangi hisseyi seçerdin?”
Dario şöyle diyor:
“Bu soruyu cevaplamamalıyım, çünkü çok fazla şey biliyorum.”
Ancak sektör olarak biyoteknoloji diyor.
Ana noktaları şöyle özetliyor:
- Biyoteknoloji, AI sayesinde büyük bir rönesans yaşayacak.
- Birçok hastalığı tedavi edebileceğiz.
- Büyük ilaç şirketleri mi yoksa küçük yeni biyoteknoloji şirketleri mi diye de tercih yapmıyor.
Bana göre bu sektörün liderlerinden biri de $TEM olacak.
Bugünü değil, geleceği satın alın.
Two weeks in the SpaceX S-1… here are the parts that took me a little longer to get comfortable with.
No drama — just transparency.
I’m still not participating (merger view unchanged), but I thought others might find it useful. Article below
🚀
During Ashok Elluswamy, VP of AI at Tesla's presentation at CVPR, a slide was shown where Tesla FSD is currently pending regulatory approval:
• All of Europe
• Japan
• India
• Israel
• Bangladesh
• Thailand
• Taiwan
• Malaysia
• Philippines
• Saudi Arabia
• UAE
• Ethiopia
• Colombia
• Chile
Combined, ~36% of the world’s population lives in these places. FSD is going global.
AI COMPUTE is sold out until end of 2028.
This is exactly how you become a millionaire, you buy and hold these companies:
GPU / CHIPS (Direct Compute Suppliers)
$NVDA
GPU monopoly. Every data center that gets built needs their chips. Sold-out compute = perpetual backlog. Pricing power is absolute.
$AMD
#2 GPU for AI inference. Overflow demand from $NVDA constraints flows here. MI300X ramp continues.
$AVGO
Custom ASIC chips (TPUs for $GOOG, XPUs for $META). Anthropic just expanded its partnership with Google to secure TPU chips supplied by Broadcom, adding multiple gigawatts of compute capacity starting in 2027. Direct beneficiary.
$MRVL
Custom AI accelerators + data center networking silicon. Wins as hyperscalers diversify away from pure $NVDA dependency.
MEMORY (You Can't Run Compute Without It)
$MU
HBM3E memory is strapped to every H100/B200. Semiconductor lead times hit 40 weeks in March 2026, with memory ICs among the most acutely constrained categories. Micron is structurally undersupplied.
$SNDK
NAND flash for data center storage. Constrained supply + surging demand = pricing power recovery.
NETWORKING (Moving Data Between GPUs)
$CRDO
High-speed SerDes and AEC cables. The connective tissue between GPUs inside clusters. Hyperscaler capex = direct revenue.
$ANET
Ethernet switching for AI clusters. As GPU clusters scale, Arista's switches are the backbone. Recurring upgrade cycle locked in.
$LITE
Optical components for data center interconnects. Fiber optic components are among the most acutely constrained categories being consumed by AI buildout. Accuris
$COHR
Transceiver modules for hyperscale networking. Same optical bottleneck play as LITE.
POWER (The Real Bottleneck)
$VST
Nuclear + natural gas power. U.S. data center electricity demand could drive price hikes of up to 79% in areas like Texas by 2027. Vistra owns generation assets in Texas. Pricing power is structural. CIO Dive
$CEG
Nuclear baseload power. Signed data center deals directly with hyperscalers. Clean, always-on power = premium pricing.
$NEE
NextEra announced plans to build a data center powered by natural gas with carbon capture tech and 1.2 GW of power capacity. Direct AI infrastructure play. Wikipedia
$GEV
Gas turbines + grid infrastructure. Every new data center needs grid upgrades. Massive backlog through 2027+.
$ETN
Power management and electrical infrastructure. Every data center that gets built runs through Eaton's switchgear and UPS systems.
DATA CENTER BUILDERS / OPERATORS
$EQIX
Largest global colocation operator. Data center space scarce through at least 2027 = pricing power for every square foot Equinix owns. KKR
$DLR
REIT with global data center footprint. Occupancy stays pinned near 100% in this supply-constrained environment.
$APLD
Pure-play AI data center builder. Just signed a $5 billion, 15-year lease with a hyperscaler for 200MW of AI/HPC infrastructure. Pipeline is locked. sec
$IREN
High-performance compute infrastructure. Pivoting from Bitcoin mining to AI GPU hosting. Low-cost power + existing infrastructure.
CLOUD HYPERSCALERS (Resell the Compute)
$MSFT
Plans to increase total AI capacity by over 80% and roughly double its data center footprint over the next two years. Azure is the distribution layer for compute. CIO Dive
$GOOG
$175–185B in capex guided for 2026. GCP is capacity-constrained. Every dollar of sold-out GPU time flows through Google Cloud. Futurum Group
$AMZN
$200B in capex projected for 2026. AWS is the dominant cloud. Sold-out compute environments compress margins upward. Futurum Group
$META
Targeting over 10 GW of total capacity by end of 2026 with capex exceeding $100B. Trains its own models in-house. Vertically integrated AI compute.
♻️ RESHARE this post and write 1 comment, I'll DM you my favorite 3 which hasn't exploded yet.
🚨 Tesla FSD is waiting for regulatory approval across a huge part of the world
During Ashok Elluswamy’s presentation at CVPR, Tesla showed pending approvals in:
• Europe
• Japan 🇯🇵
• India 🇮🇳
• Israel 🇮🇱
• Bangladesh 🇧🇩
• Thailand 🇹🇭
• Taiwan 🇹🇼
• Malaysia 🇲🇾
• Philippines 🇵🇭
• Saudi Arabia 🇸🇦
• UAE 🇦🇪
• Ethiopia 🇪🇹
• Colombia 🇨🇴
• Chile 🇨🇱
These regions represent roughly 36% of the world’s population. $TSLA
You got $100,000 to invest
Which stock makes you a millionaire by 2028?
1. $AMD
2. $NBIS
3. $MU
4. $META
5. $PLTR
6. $MSFT
7. $NOW
8. $IREN
9. $MSTR
10. $INTC
England is building one of the largest battery gigafactories in Europe.
Somerset. Bridgwater. 4,200 direct jobs. £380 million government investment. Built with 100% British steel.
It's expected to generate £43 billion in economic growth over 25 years and supply batteries for JLR.
Despite how it sometimes feels, England isn't done making things.
Not even close.
#MadeInEngland
More and more are realising that FF's are responsible for the high price of electricity in the UK, along with an outdated pricing mechanism. Something I've been saying for a number of years.
@LoftusSteve will continue to deny it, but the inescapable fact remains. Whether you now include LCOE (essentially Capex, ops cost, depreciation and decommissioning) or not, renewables come out cheapest. I've used Groks summary's to help simplify.
The deniers will continue to deny, whilst those who embrace renewables continue to benefit from lower energy costs. the truth is in the wallets of those who know.
https://t.co/HxdXlMsdNL
This is insane...
It took 69 years for governments all around the world to launch nearly 4,900 Stateline in orbit.
But it only took SpaceX 8 years to launch nearly 10,000.
🤯
Tesla's FSD 15 is compatible with AI4, Cybercab production is already ramping, and the safety data is there. @TashaARK, CFA shares why she's confident AI4 is the platform that will deliver robotaxi in our May webinar.
@1RustyMac@PeterDiamandis I've talked about the impact of mass adoption theory and crossing the chasam many times over the years.
I even mapped it for everyone a very long time ago:
@AllQuizAble@KateFantom ICE as a technology won't die, same as Steam & Horses are still around.
However, ICE for transportation is now well on its decline - Big Oil is saying sales of fossil fuel are dropping year on year.
Good luck refuelling your diesel in a few short years...
Getting positioned in these stocks before June 15 could be a life-changing move.
1. $ALAB - Astera Labs
2. $VRT - Vertiv Holdings
3. $CRWD - CrowdStrike Holdings
4. $CRWV - CoreWeave
5. $NVDA - NVIDIA
6. $PLTR - Palantir Technologies
7. $TSLA - Tesla
8. $RKLB - Rocket Lab USA
Don't overlook this trend. Jensen Huang is betting on it. Elon Musk has been building for it for years. The investors who recognize the opportunity early may have the most to gain as this industry continues to scale.
Elon Musk's children don't go to normal school. And the reason why will change how you think about education.
He pulled his kids out of one of the most prestigious schools in Los Angeles. Parents were furious. Media called him arrogant. The school had a waitlist of thousands.
His response: "They're teaching kids to solve problems that already have answers. I need them to solve problems nobody's thought of yet."
So he built a school. Inside SpaceX. Called it Ad Astra. No grades. No tests. No subjects in the traditional sense.
A nine year old could take apart a rocket engine and present their findings to actual SpaceX engineers. Students didn't study history. They debated whether they'd make different decisions than historical leaders using the same information available at the time.
The school had no grade levels. A seven year old could work alongside a thirteen year old if they were interested in the same problem.
When asked why he structured it this way, Elon said something that stuck with me:
"I don't care if they know the answer. I care if they know which questions are worth asking."
Most people spend their entire education learning how to be right. Elon teaches his children how to be curious.
The system rewards answers. Life rewards questions.