自由1688
284 Posts
自由1688
ETH Holder
Occasional Trader
1.9 Years
559 Following
22.4K+ Followers
13.8K+ Liked
Posts
All
Quotes
Videos
Live
PINNED
PINNED
🧧🧧🧧 Bonus Time! 🧧🧧🧧
👉👉👉 Follow & Comment 👈👈👈
🔥🔥🔥 To claim your cash bonus! 🔥🔥🔥
👉👉👉 Follow & Comment 👈👈👈
🔥🔥🔥 To claim your cash bonus! 🔥🔥🔥
🚨 A rampaging bull market—airdrops straight to the face! Click in to check the rules and use your speed to claim an exclusive cash red envelope! 🧧
自由1688
·
--
🧧🧧🧧 Bonus Time! 🧧🧧🧧
👉👉👉 Follow & Comment 👈👈👈
🔥🔥🔥 To claim your cash bonus! 🔥🔥🔥
👉👉👉 Follow & Comment 👈👈👈
🔥🔥🔥 To claim your cash bonus! 🔥🔥🔥
🧧⚠️ Warning: Chip inventory is running out! Big companies are疯狂ly stocking up. With compute costs skyrocketing, how should we plan our Web3 strategy?
自由1688
·
--
Storage chips are facing a full-blown crisis! Samsung & SK hynix inventory is under 10 days—will AI infrastructure be bottlenecked again?
Brothers, I just saw a hard-hitting report released by South Korean brokerage KB Securities. The AI hardware “foundation” is about to make news again! The shortage of storage semiconductors hasn’t eased at all—it’s actually deteriorating rapidly.
Directly highlighting the key points for everyone:
1️⃣ Inventory is critically low: Samsung Electronics and SK hynix’s inventory of memory chips has already dropped to less than 10 days of supply! With the疯狂扩张 of investments in AI infrastructure, next year will very likely see a serious shortage of supply versus demand. 2️⃣ HBM4 as the “capacity-eating beast”: The industry is moving toward the next-generation high-bandwidth memory (HBM4), but it requires wafers that are 3 times that of traditional DRAM. Once HBM4 is fully mass-produced, the capacity of traditional DRAM will be cut significantly. Institutions estimate that next year demand for DRAM and NAND will exceed supply by more than 10%. 3️⃣ An epic shortage wave: It’s not just HBM—AI servers will also consume massive amounts of server DDR5 memory and enterprise-grade solid-state drives (SSDs). 4️⃣ Big players are throwing money around: Global data center giants have raised their expected investments in AI infrastructure next year to $1.3 trillion (up 60% year over year). Moreover, the share of storage semiconductors in AI infrastructure investment is expected to jump from 14% last year to 57% (and some institutions even predict as high as 68%)!
Directly highlighting the key points for everyone:
1️⃣ Inventory is critically low: Samsung Electronics and SK hynix’s inventory of memory chips has already dropped to less than 10 days of supply! With the疯狂扩张 of investments in AI infrastructure, next year will very likely see a serious shortage of supply versus demand. 2️⃣ HBM4 as the “capacity-eating beast”: The industry is moving toward the next-generation high-bandwidth memory (HBM4), but it requires wafers that are 3 times that of traditional DRAM. Once HBM4 is fully mass-produced, the capacity of traditional DRAM will be cut significantly. Institutions estimate that next year demand for DRAM and NAND will exceed supply by more than 10%. 3️⃣ An epic shortage wave: It’s not just HBM—AI servers will also consume massive amounts of server DDR5 memory and enterprise-grade solid-state drives (SSDs). 4️⃣ Big players are throwing money around: Global data center giants have raised their expected investments in AI infrastructure next year to $1.3 trillion (up 60% year over year). Moreover, the share of storage semiconductors in AI infrastructure investment is expected to jump from 14% last year to 57% (and some institutions even predict as high as 68%)!
💰 Money-throwing at the square! Cash blind boxes are ready—just two steps (follow + comment), and it’s credited instantly! 🧧
自由1688
·
--
🧧🧧🧧 Bonus Time! 🧧🧧🧧
👉👉👉 Follow & Comment 👈👈👈
🔥🔥🔥 To claim your cash bonus! 🔥🔥🔥
👉👉👉 Follow & Comment 👈👈👈
🔥🔥🔥 To claim your cash bonus! 🔥🔥🔥
🧧🤯 HBM4 becomes a “capacity-consuming monster”! AI infrastructure is once again being held back—how will this hardware shortage affect Crypto?
自由1688
·
--
Storage chips are facing a full-blown crisis! Samsung & SK hynix inventory is under 10 days—will AI infrastructure be bottlenecked again?
Brothers, I just saw a hard-hitting report released by South Korean brokerage KB Securities. The AI hardware “foundation” is about to make news again! The shortage of storage semiconductors hasn’t eased at all—it’s actually deteriorating rapidly.
Directly highlighting the key points for everyone:
1️⃣ Inventory is critically low: Samsung Electronics and SK hynix’s inventory of memory chips has already dropped to less than 10 days of supply! With the疯狂扩张 of investments in AI infrastructure, next year will very likely see a serious shortage of supply versus demand. 2️⃣ HBM4 as the “capacity-eating beast”: The industry is moving toward the next-generation high-bandwidth memory (HBM4), but it requires wafers that are 3 times that of traditional DRAM. Once HBM4 is fully mass-produced, the capacity of traditional DRAM will be cut significantly. Institutions estimate that next year demand for DRAM and NAND will exceed supply by more than 10%. 3️⃣ An epic shortage wave: It’s not just HBM—AI servers will also consume massive amounts of server DDR5 memory and enterprise-grade solid-state drives (SSDs). 4️⃣ Big players are throwing money around: Global data center giants have raised their expected investments in AI infrastructure next year to $1.3 trillion (up 60% year over year). Moreover, the share of storage semiconductors in AI infrastructure investment is expected to jump from 14% last year to 57% (and some institutions even predict as high as 68%)!
Directly highlighting the key points for everyone:
1️⃣ Inventory is critically low: Samsung Electronics and SK hynix’s inventory of memory chips has already dropped to less than 10 days of supply! With the疯狂扩张 of investments in AI infrastructure, next year will very likely see a serious shortage of supply versus demand. 2️⃣ HBM4 as the “capacity-eating beast”: The industry is moving toward the next-generation high-bandwidth memory (HBM4), but it requires wafers that are 3 times that of traditional DRAM. Once HBM4 is fully mass-produced, the capacity of traditional DRAM will be cut significantly. Institutions estimate that next year demand for DRAM and NAND will exceed supply by more than 10%. 3️⃣ An epic shortage wave: It’s not just HBM—AI servers will also consume massive amounts of server DDR5 memory and enterprise-grade solid-state drives (SSDs). 4️⃣ Big players are throwing money around: Global data center giants have raised their expected investments in AI infrastructure next year to $1.3 trillion (up 60% year over year). Moreover, the share of storage semiconductors in AI infrastructure investment is expected to jump from 14% last year to 57% (and some institutions even predict as high as 68%)!
Apple is brewing its biggest change since the iPhone X! “iPhone 20” preview revealed 📱
According to the latest supply chain news, Apple plans to launch the highly disruptive “iPhone 20” in 2027 (the 20th anniversary of the iPhone), which could be the most radical redesign since the iPhone X in 2017.
💡 Core upgrade highlights:
- Seamless glass aesthetics: Say goodbye to right-angle bezels and adopt “equal-depth quad-curved glass,” fully moving toward the minimalist form of “a single seamless piece of glass,” with an almost borderless look.
- Eliminating physical buttons: Pressure-sensitive solid-state buttons will replace all mechanical buttons (volume, power, etc.), combined with vibration motors, ending wear on mechanical cutouts!
- Racing toward a true full-screen display: Apple is developing a solution to completely hide Face ID and the front-facing camera under the display, achieving a front side with “zero cutouts.”
- Hardcore AI and computing power: It is expected to be powered by the A21 chip built on TSMC’s 2nm process, and may for the first time introduce an HBM (high-bandwidth memory) architecture dedicated to on-device large models, along with Apple’s first self-developed 5G baseband and a new pure-silicon battery, fully maximizing both AI performance and battery life!
Opinion: Apple’s major moves in hardware and on-device AI computing power will not only determine the form of smartphones for the next few years, but also provide a stronger foundation for the development of Web3, on-device AI applications, and crypto hardware wallets.
What do you think—can the iPhone 20, with this major redesign, change the world again? 👇 #Apple #iPhone20 #端侧AI #Web3硬件
According to the latest supply chain news, Apple plans to launch the highly disruptive “iPhone 20” in 2027 (the 20th anniversary of the iPhone), which could be the most radical redesign since the iPhone X in 2017.
💡 Core upgrade highlights:
- Seamless glass aesthetics: Say goodbye to right-angle bezels and adopt “equal-depth quad-curved glass,” fully moving toward the minimalist form of “a single seamless piece of glass,” with an almost borderless look.
- Eliminating physical buttons: Pressure-sensitive solid-state buttons will replace all mechanical buttons (volume, power, etc.), combined with vibration motors, ending wear on mechanical cutouts!
- Racing toward a true full-screen display: Apple is developing a solution to completely hide Face ID and the front-facing camera under the display, achieving a front side with “zero cutouts.”
- Hardcore AI and computing power: It is expected to be powered by the A21 chip built on TSMC’s 2nm process, and may for the first time introduce an HBM (high-bandwidth memory) architecture dedicated to on-device large models, along with Apple’s first self-developed 5G baseband and a new pure-silicon battery, fully maximizing both AI performance and battery life!
Opinion: Apple’s major moves in hardware and on-device AI computing power will not only determine the form of smartphones for the next few years, but also provide a stronger foundation for the development of Web3, on-device AI applications, and crypto hardware wallets.
What do you think—can the iPhone 20, with this major redesign, change the world again? 👇 #Apple #iPhone20 #端侧AI #Web3硬件
🚀 Tesla Cybercab officially starts carrying passengers! The era without steering wheels is here—will we still need to buy cars in the future?
Tesla’s autonomous driving grand vision has taken another historic step! On September 3, the Cybercab, which completely eliminates the steering wheel and brake pedal, officially joined Tesla’s Robotaxi network and began real passenger service in Austin, Texas.
💡 Key highlights at a glance:
- Disruptive design: Built specifically for driverless operation, it comes with cool butterfly doors. The cabin retains a large screen, and with the Starlink network, passengers can even watch 4K videos smoothly in the car, like a “mobile lounge.”
- An extreme cost killer: Musk’s goal is to cut operating costs to 20 cents per mile, while charging passengers only 30–40 cents per mile. If this ultra-low-price model succeeds, cars in the future may completely shift from “private assets” to “on-demand mobility services,” and buying a car may no longer be a necessity!
- Market and challenges: Austin currently has 45 registered Cybercabs on the road. Although it still faces regulatory review from the National Highway Traffic Safety Administration (NHTSA) and the scale remains limited, this move has become a core factor in the capital markets’ valuation of Tesla ($TSLA)’s future prospects.
From ride-hailing platforms to autonomous fleets, Musk’s AI and mobility strategy is accelerating into reality. Do you think Cybercab can truly replace private cars? Feel free to share your thoughts in the comments! 👇
#科技前沿 #自动驾驶 #Tesla
Tesla’s autonomous driving grand vision has taken another historic step! On September 3, the Cybercab, which completely eliminates the steering wheel and brake pedal, officially joined Tesla’s Robotaxi network and began real passenger service in Austin, Texas.
💡 Key highlights at a glance:
- Disruptive design: Built specifically for driverless operation, it comes with cool butterfly doors. The cabin retains a large screen, and with the Starlink network, passengers can even watch 4K videos smoothly in the car, like a “mobile lounge.”
- An extreme cost killer: Musk’s goal is to cut operating costs to 20 cents per mile, while charging passengers only 30–40 cents per mile. If this ultra-low-price model succeeds, cars in the future may completely shift from “private assets” to “on-demand mobility services,” and buying a car may no longer be a necessity!
- Market and challenges: Austin currently has 45 registered Cybercabs on the road. Although it still faces regulatory review from the National Highway Traffic Safety Administration (NHTSA) and the scale remains limited, this move has become a core factor in the capital markets’ valuation of Tesla ($TSLA)’s future prospects.
From ride-hailing platforms to autonomous fleets, Musk’s AI and mobility strategy is accelerating into reality. Do you think Cybercab can truly replace private cars? Feel free to share your thoughts in the comments! 👇
#科技前沿 #自动驾驶 #Tesla
Barren land blossomed with dazzling flowers
If you made money in the crypto world, would you help these children?
#beauty #rural #left-behind children
If you made money in the crypto world, would you help these children?
#beauty #rural #left-behind children
A group of married men got together for a trip and shared the “boring” process,
Netizens: If you film one more second, you can’t hold back the corners of your mouth
$NVDA.US 😍🚀
Netizens: If you film one more second, you can’t hold back the corners of your mouth
$NVDA.US 😍🚀
After ETH surged more than 20%, “7 Siblings” sells 9,000 ETH to obtain 21.04 million USDT
On August 21 (UTC+8), according to on-chain analyst Yu Jin’s monitoring, a mega whale/institution named “7 Siblings” bought after ETH fell by more than 10% in a short period, and sold after ETH rose by more than 10% in a short period. Earlier this year, in February and June, after ETH saw two instances of declines of more than 10%, this whale/institution also made buys; the buy price in June was $1,789. Over the past day and a half, ETH has risen by more than 20%, and 7 Siblings (0x28a...a6b) has started selling. Within the past 6 hours, it sold 9,000 ETH, exchanging it for 21.04 million USDT, with an average price of $2,338. (Source: ODAILY)
On August 21 (UTC+8), according to on-chain analyst Yu Jin’s monitoring, a mega whale/institution named “7 Siblings” bought after ETH fell by more than 10% in a short period, and sold after ETH rose by more than 10% in a short period. Earlier this year, in February and June, after ETH saw two instances of declines of more than 10%, this whale/institution also made buys; the buy price in June was $1,789. Over the past day and a half, ETH has risen by more than 20%, and 7 Siblings (0x28a...a6b) has started selling. Within the past 6 hours, it sold 9,000 ETH, exchanging it for 21.04 million USDT, with an average price of $2,338. (Source: ODAILY)
Alibaba AI model downloads surpass 3 billion times
In the past six months, the global cumulative download count of the open-source models under Alibaba has exceeded 3 billion times, surpassing U.S. tech giant Meta and Google, as well as Chinese peers, making it the AI model with the highest number of downloads worldwide. Alibaba said that its AI model series, Qwen, has been open-sourced with more than 460 models, and that its ecosystem includes over 300,000 derived models.
A report released by the open-source AI platform Hugging Face on August 14 shows that this year Google’s model downloads reached 418 million times, while Meta’s model downloads were 227 million. The report states that Qwen’s download data makes it one of the largest foundations of the openAI ecosystem: “For developers who are deciding which models to fine-tune and deploy, Qwen has become part of their default workflow.”
— Lianhe Zaobao, Bloomberg
In the past six months, the global cumulative download count of the open-source models under Alibaba has exceeded 3 billion times, surpassing U.S. tech giant Meta and Google, as well as Chinese peers, making it the AI model with the highest number of downloads worldwide. Alibaba said that its AI model series, Qwen, has been open-sourced with more than 460 models, and that its ecosystem includes over 300,000 derived models.
A report released by the open-source AI platform Hugging Face on August 14 shows that this year Google’s model downloads reached 418 million times, while Meta’s model downloads were 227 million. The report states that Qwen’s download data makes it one of the largest foundations of the openAI ecosystem: “For developers who are deciding which models to fine-tune and deploy, Qwen has become part of their default workflow.”
— Lianhe Zaobao, Bloomberg
Log in to explore more content