percccca

[NIK @ns123abc]

🚨BREAKING: SoftBank tried to borrow $6 billion against its 13% OpenAI stock… to keep funding OpenAI

Banks said NO

They don’t believe OpenAI is worth $852 billion 💀

[percccca @parccccccc]

@dameanonyme เจ้าสาวเก่งมากเลยนะคะ เดินขึ้นไปได้ทั้งชุดโดยเหงื่อไม่แตกก่อน แถมตรงนั้นตอนไปไม่กล้าเข้าห้องน้ำเลย เอเวอรี่ติงดู medieval สุดๆเลยย

[Damnang2 @damnang2]

https://t.co/XZ7HHl1uiD

[percccca @parccccccc]

@letstalkaboutKR เจนเซ่น หวง* ค่า(JENSEN HUANG)

[percccca @parccccccc]

@minminnayy ยังใช้ ญป ได้หรอคะจำกัดระยะทางไว้เยอะมากเลย จ่ายเพิ่มก็แพงมาก เผลอๆไปพวกแพลตฟอร์มยังถูกกว่าจ่ายระยะทางเพิ่มจากที่กำหนด

[percccca @parccccccc]

@JOMMJOM @Bonusiie คิดว่าที่คนห่วงกันน่าจะถ้าเป็น เพราะดูเหนื่อยง่ายขึ้นเยอะมาก ถ้ากรณีเข้าคลีนิคก็อยากให้ค���ณเอิ้กไปตรวจสุขภาพที่รพ.ควบคู่ไปด้วย การที่เปลี่ยนแปลงเร็วขนาดนี้ เป็นห่วงว่าค่าต่างๆจะอยู่ในระดับปกติไหม แต่ถ้าลดแบบค่อยเป็นค่อยไปมีแพทย์ดูแลใกล้ชิด เช็คค่าต่างๆอย่างสม่ำเสมอก็ดี

[CG @cgtwts]

> be Yann LeCun
> spend years building JEPA at Meta
> company focuses on LLaMA instead
> his idea stays complicated and unused
> robotics plans get dropped
> decides to leave and start AMI Labs
> builds a much simpler version from scratch
> trains it on normal hardware in just a few hours
> removes all the complicated tricks and keeps it simple

Results:

-uses 200x less data than similar systems
-makes decisions 50x faster
-runs on a single GPU instead of massive clusters
-simple to train
-understands movement, objects, and space
-can tell when something is physically impossible
-learns how the real world works without being explicitly taught.

[Isabella Maria DeLuca @IsabellaMDeLuca]

Lululemon, along with other legging brands, have forever chemicals in their clothes.

These chemicals don’t just wash out. They don’t break down. And once they’re in your body— they don’t leave.

These can be absorbed through the skin, which makes it worse when you learn that these forever chemicals are heavily concentrated in the crotch region of leggings. The most absorbent, sensitive skin on your body. Worn tight. For hours. Sometimes with no barrier.

PFAS are linked to: hormone disruption, thyroid issues, live and kidney damage, increased cancer risk, and infertility.

I hate it here.

[จุ้งเทมปุระเหนื่อยๆ @goodforsoul16]

สูตรจ้า เผื่อทำกินวันสงกรานต์ ปรับสูตรมาจากสูตรข้าวอบเห็ดกรอบมิโสะร้านโมโมฟุกุที่เป็นไวรัลอยู่ช่วงนึง ปรับสูตรแบบไม่ใช้เตาอบ+เครื่องปรุงเท่าที่มี

เครื่องปรุง สำหรับ 4 ที่
-ข้าวญี่ปุ่นหรือถ้าใครซื้อข้าวโดยบือโปะโหละโพปรีไปก็ใช้ได้ 1.5ถ้วยตวง ซาวให้สะอาด แช่น้ำไว้ 30นาทีขึ้นไป

[จุ้งเทมปุระเหนื่อยๆ @goodforsoul16]

ตั้งแต่เริ่มทำข้าวอบเห็ดก็คือมีสั่งกันจะสิบหม้อแล้ว เลิกทำขนมมาขายข้าวอบเห็ดดีมั๊ย แง้ จริงๆมันทำง่ายมาก

[Trade Whisperer @TradexWhisperer]

Having worked in the memory industry for 20 years. I respectfully disagree with @jukan05 on HBM3E

Late qual means late engineering learning. Every wafer SK Hynix and Micron ran while Samsung was still in retest cycles was compounding process knowledge Samsung doesn't have yet.

Samsung failed NVIDIA HBM3E qual for roughly 18 months. The root cause wasn't throughput. It was thermal and TSV performance. That's a process maturity problem, not a capacity problem. You can output more wafers at lower yields and provide the supply but at a cost.

At 12-hi stacking, yield isn't linear. Each additional die multiplies defect probability. A 0.5% per-die yield gap compounds across 12 layers. DPM and leakage issues don't disappear with volume. They take time in production to diagnose, redesign, and fix.

Samsung held 41% of the HBM market in Q2 2024. By Q2 2025 they were at 17%. Micron went from 4% to 21% in the same window. That's what late HBM3E qual costs you.

That answer was SK Hynix first. Micron second. Samsung third. HBM4 might be a different story. We don't know yet.

Yield curves do converge but the race for highest HBM3E 12-hi maturity yield and reliability will continue on for years, especially due to exponentially higher complexity of high bandwidth memory.

I've seen this over and over again through decades of my career with different companies.

$MU $DRAM

[Halal Investor @_HalalTrader_]

$MU “We expect very aggressive buybacks starting Dec 2026.”

“No impact from the war in Iran. Micron is in a better position than Korean players from a Helium, LNG and raw materials standpoint as its supply chain is domestic to US.”
-Cantor

Looking at the cash flow Micron can deploy, this should fly starting December InshaAllah.

Cantor reiterates PT of 700 after meeting with the CEO and exec team.

[Jukan @jukan05]

Views on Memory Stock Price Decline
[Samsung Securities Semiconductor, IT / Jong-wook Lee]

As of 10:30 AM, Samsung Electronics and SK Hynix shares are down 3.9% and 5.6%, respectively.

1. Spot Price Decline

D5 16G spot prices peaked on March 19 and have since corrected over the past week, with the premium over contract prices narrowing from a 1Q average of 30% to roughly 20% recently. The equity market is concerned this may mark an inflection point signaling DRAM price declines.

We believe the pace of DRAM price increases will begin to moderate from 3Q onward, and that early signs of this deceleration could trigger a share price correction around May–June. From this same perspective, the narrowing of spot price premiums appears to be interpreted as evidence of slowing DRAM price momentum.

2. Views on the Correction

We believe this is a correction, not a peak-out. The market narrative will shift from the magnitude of DRAM price increases to the sustainability of DRAM prices. Because it is difficult to pinpoint the exact inflection between these two themes, we believe riding through this correction offers a more favorable risk-return profile.

We see a high probability that DRAM profits will trace a broad plateau rather than the typical cyclical peak, as the supply-side response has lagged roughly one year behind compared to previous cycles. We believe this plateau-shaped earnings profile will instill greater confidence among investors in the sustainability of the cycle.

3. The Cycle

We maintain that DRAM remains a cyclical industry. That said, we caution against selling on the vague fear that "it's cyclical, so it has to come down eventually."

Cyclical peak-outs originate from changes in downstream industries. Therefore, early signs of a peak-out can be identified through business trajectory shifts at OpenAI, Anthropic, and the ISPs — and we believe there is still upside to be demonstrated from the demand side.

Looking back at 2025, the equity market has wobbled every time spot prices stalled. While spot price declines can induce fear among investors, we do not believe they should be extrapolated as evidence of a cyclical peak-out.

4. Natural Gas Supply Issues

Natural gas is a cost inflation issue, not a production disruption issue. In memory, cost is dominated by the variability of technology, yields, and capex — which has historically overwhelmed raw material cost fluctuations. Moreover, in a commodity industry, raw material supply disruptions cannot be viewed as purely negative. What matters is whether demand is strong enough to absorb cost pass-throughs.

5. Sentiment

Examining the drivers behind the recent share price correction, this is a sentiment issue rather than a fundamental one — though sentiment itself is a meaningful factor in equity markets. Investors are feeling significant anxiety about sustainability following the sharp run-up in both DRAM prices and stock prices, and they want concrete evidence to alleviate that anxiety.

If tangible evidence emerges regarding the direction from 3Q onward, we believe share prices can stabilize and begin to reflect earnings with greater confidence.

Thank you.

(Published 2026/3/30)

[percccca @parccccccc]

@LittleBirbMame มูหูอย่างเดียว ชื่อ ginza onodera แต่อยู่ omotesando! 555

[bubble boi @bubbleboi]

If this paper is true (I have no doubt it is) this has massive implications on memory supply and yields.

DRAM up to know used a 6F² cell layout. Moving to 4F² means the cell area shrinks to ~67% of the conventional size. They are able to do this with a “vertical channel” where transistors are upright rather than flat. By also stacking the peripheral circuitry for controlling the memory cells they get better density too. TLDR it’s a sky scraper rather than a large foot print building.

So basically this means they get 20% more DRAM dies per wafer without changing the node or process.

Absolutely wild.

[bone @boneGPT]

DEVELOPING: Chinese entrepreneur boasts receipt of 200 NVIDIA H200 GPUs in Beijing despite US export ban, explains how he circumvents export ban 🧵

[Mark Gadala-Maria @markgadala]

We have reached peak AI.

Someone made an entire short film "Harry Potter by Balenciaga"

Credit: demonflyingfox on YT

[Trade Whisperer @TradexWhisperer]

https://t.co/hVDXWL3v9k

[Daniel Romero @HyperTechInvest]

$MU enters high-volume production of HBM4 for NVIDIA Vera Rubin

Micron said it has started volume shipments of its 36GB 12-high HBM4 in the first quarter of 2026, and the product is designed for NVIDIA Vera Rubin systems

The company says the chip delivers over 11 Gb/s pin speed, more than 2.8 TB/s of bandwidth, around 2.3x the bandwidth of HBM3E, and over 20% better power efficiency

Micron also said it has shipped samples of its 48GB 16-high HBM4, which increases capacity per HBM placement by 33% versus the 36GB 12-high version

Beyond HBM, Micron said its 192GB SOCAMM2 is now in high-volume production. It is designed for NVIDIA Vera Rubin NVL72 systems and standalone NVIDIA Vera CPU platforms, enabling up to 2TB of memory and 1.2 TB/s of bandwidth per CPU

Micron also said it is the first company to mass-produce a PCIe Gen6 data center SSD. Its Micron 9650 is built for AI training and inference workloads, supports up to 28 GB/s sequential read throughput and 5.5 million random read IOPS, and is optimized for NVIDIA BlueField-4 STX architecture. The company says it offers up to 2x the read performance of Gen5 and 100% higher performance per watt

[percccca @parccccccc]

@chudalee @bongtao @soyoso หลังๆทาน เครื่องเยอะเหมือนเดิม แต่เหมือนคุณภาพหมูเด้งไม่เท่าเดิม แป้งเยอะมาก แต่ก็ต้นทุนหมูขึ้นมาเยอะจริง

ส่วนเส้นญวนเราเคยสั่งสองสาขา(คนละที่)เหม็นอับเหมือนกัน🫠