ReadWriteBeHappy

Working on autonomous driving in Korea !
SW engineer, amateur boxing, hiking.
Mainly retweet ML / CV related things. Invest in AI. Personel views only!!! ♥

[ReadWriteBeHappy @treasured_write]

📚 2026년 독서 타래

[NVIDIA DRIVE @NVIDIADRIVE]

We are helping autonomous vehicle researchers go beyond recorded miles.

At #CVPR2026, we unveiled new AV agent skills that help turn fleet data into editable 3D scenes, generate synthetic long-tail driving scenarios and evaluate policies in closed-loop simulation.

With Neural Reconstruction skills, NVIDIA AlpaGym and NVIDIA Alpamayo 2 Super, researchers can iterate across perception, planning and action for safer, scalable L4 development.

Learn more: https://t.co/JUBRFBnYeH

[Reuters @Reuters]

Nvidia CEO Jensen Huang said the company has enough supply to accommodate robust growth in central processing units and graphics processing units as it rides an AI boom https://t.co/UsK82dFXig

[tae kim @firstadopter]

An NVIDIA executive just told me the GPU-to-CPU ratio will go from 2:1 today to 1:1 “in months" due to agentic AI.

NVIDIA executive is saying 1:1 IN MONTHS.

CPU CPU CPU $NVDA

[Niels Rogge @NielsRogge]

@NASNETou Find all CVPR papers here: https://t.co/jD01frdcRF

[FP Blackbird @Frederi55284379]

$SAP ils n'ont jamais eu autant de clients...le Marché va réajuster.

[日本経済新聞 電子版（日経電子版） @nikkei]

ポケモンの純利益が71%増の1200億円　30周年効果、26年2月期
https://t.co/8D1qTRcQ0D

[OpenAI @OpenAI]

OpenAI frontier models and Codex are now generally available on AWS, giving enterprises a new way to build on Amazon Bedrock with OpenAI through the security, compliance, and governance workflows they already use.

This is also the beginning of a broader expansion of OpenAI capabilities on AWS, including future availability for cybersecurity capabilities like Daybreak.

https://t.co/vMws0YU6Q3

[Jukan @jukan05]

Kioxia seems to believe that although the supply-demand curve will gradually recover, NAND will still remain in shortage even in Q4 2027.

[tae kim @firstadopter]

Very very interesting. Foxconn going to partner with Intel to make rackscale CPU racks (Intel Xeon)

[NVIDIA DRIVE @NVIDIADRIVE]

What if you could hear what your car is thinking? 💭

This NVIDIA Alpamayo demo gives voice to the internal reasoning behind an autonomous drive — showing how the vehicle evaluates the road, plans its next move and adapts to city traffic while the driver enjoys the ride.

Watch the demo 👉 https://t.co/8SsRriGdwu

[Boris Cherny @bcherny]

Claude Opus 4.8 is out today. It's our strongest coding model yet: up on SWE-bench Pro (from 64.3 to 69.2) and noticeably more honest about its own work. It tells you when it's unsure and catches its own bugs instead of declaring victory early. Same price as 4.7.

[Rekt Fencer @rektfencer]

UPDATE

Leopold Aschenbrenner just increased one stock position by 650% in a single quarter.

While everyone was looking at his $8.5B in semiconductor puts, his real conviction bet was under the radar.

It is $CLSK.

$BTC miner transitioning into AI data center infrastructure.

He isn't chasing chips.

He is buying what AI will need when the real bottleneck hits.

[tae kim @firstadopter]

Nvidia's new Vera CPU kicks ass $NVDA

"NVIDIA Vera CPU Benchmarks: Olympus Cores Delivering The Best Performance Ever Seen On ARM"

"On a geo mean basis, the NVIDIA Vera delivered 10% better performance than the AMD EPYC 9575F 5.0GHz high frequency processor. For gen-on-gen compared to Grace, Vera was coming in at 1.63x the performance geo mean. Over a single Intel Xeon 6980P as Intel's current flagship Granite Rapids processor, NVIDIA Vera delivered 1.55x the performance."

[木下洋介 @yosuke_kinosita]

2045年までの未来予測
つまり現状の日本は、石油が来ず債務危機が目前にあり
長期金利が暴騰し、大暴落があるかも知れないという前提にあるが

同時にAIの指数関数的成長で
今現在は何もかも設備が足りない
今現在の戦後の焼け野原状態であり、高度経済成長が目前にある状態という事。まずは大暴落かも知れないが、結局株価は意味不明な水準までぶち上るという事。日経平均は100万円を超えるって話。

失業はホワイトカラーから物質制約への移動であり、ワイの予想ではクリエイティブな仕事以外はしたくない！という失業を除き、労働需要そのものはひっ迫し続けるだろう。ロボットを動員しようが、ニートのオッサン、老人を動員しようが移民をかき集めようがロボットを十分に動かす実物資産も資源も結局何もかも足りないのだ！ロボットで失業！なんてのは遠い未来の話である。

今は戦後の焼け野原であり高度経済成長期前の日本である。働け！そして定期的に株を買うのだ！タイミングを当てるのはムズかしい。定期的に買え！銅は国家なり！

[NVIDIA DRIVE @NVIDIADRIVE]

Heading to #CVPR2026?

The Alpamayo Summit is currently at capacity, but you can still join the waitlist for a chance to attend this half-day workshop on Alpamayo, AlpaSim, the Physical AI AV dataset, and more.

Registered attendees will receive access to the replay.

Join the waitlist: https://t.co/ywWRfEiuD0

[Convequity @convequity]

800V DC Data center seems to be progressing faster than expected. IMO that's the biggest changing factor to the power theme that people don;t point out.

[Niels Rogge @NielsRogge]

One of the hottest terms in AI right now is "On-policy distillation".

It is a post-training technique in which a student model, typically an LLM, samples from its current policy and receives a teacher signal for on-policy states. It combines the dense supervision of distillation with the locality of online RL.

Now a method on PapersWithCode!

Find all 183 papers that cite it, and more here: https://t.co/NIsUjyU3UP

[aramco @aramco]

Our first-quarter performance demonstrates our resilience as a global energy leader navigating a complex geopolitical landscape, while continuing to meet the world’s energy needs

[Wall St Engine @wallstengine]

XIAOMI CEO WARNS SMARTPHONE PRICES WILL KEEP RISING

Lei Jun says smartphone prices are likely to climb over the next two years as memory chip costs rise.

The pressure is coming from AI data-center demand pulling supply toward higher-margin memory products.

TrendForce says conventional DRAM contract prices rose 90%-95% QoQ in Q1 2026.

NAND flash prices rose 55%-60% in the same period.

Xiaomi already raised prices on several handset models earlier this year by about $29.

The AI memory shortage is no longer just a data-center problem.

It is starting to show up in phones, PCs, consoles and other consumer electronics.

[Wall St Engine @wallstengine]

HUAWEI HAS A NEW PATH AROUND CHIP SCALING LIMITS

Huawei just unveiled its Tau Scaling Law, a design approach meant to reduce signal delay instead of relying only on shrinking transistors.

The key technology is called LogicFolding. https://t.co/6kW6jGq9YZ

Instead of simply making transistors smaller, Huawei says it can rearrange logic blocks to shorten wiring paths, reduce latency, improve density, and boost power efficiency.

This is Huawei’s attempt to narrow the gap with TSMC despite U.S. restrictions limiting access to advanced chipmaking tools like EUV.

Huawei says it has already designed and mass-produced 381 chips using this approach over the past six years.

Its next Kirin smartphone chip, expected this fall, will reportedly use LogicFolding and deliver:

53% higher density
41% better power efficiency
12% faster clock speeds

Huawei expects high-end chips based on Tau Scaling Law to reach transistor density equivalent to 1.4nm processes by 2031.

This does not mean Huawei has already caught TSMC or solved advanced manufacturing. It means Huawei is trying to offset process-node limits through chip architecture, circuit layout, and system-level design.

Still, if the claims hold up in mass production, this is one of the more serious attempts yet to reduce dependence on traditional Moore’s Law scaling.