𝛼

[𝛼 @deeperflows]

This is exactly why I would expect RTX 6000 PROs to reach at least $40k within 12 months. Might be wrong.

But so far, never was.

[Ayush 🙏 @ayushtweetshere]

this is Dario and sama's worst nightmare -

A law firm buying Nvidia servers.

Latham & Watkins is building an in-house AI stack..

this is US’s second-largest law firm with $8.3 Billion in revenue last year

And now it has -

- Nvidia hardware it controls
- open-weight models it can fine tune
- proprietary legal data it is trusted to protect
- infrastructure only Latham employees can access

A law firm has decades of contracts, negotiations, client context, legal reasoning, and institutional knowledge.

They dont want to give all of that away to OpenAI or Anthropic in exchange for expensive tokens..

And on top of that - risk their data being used to train frontier models..

This will happen more and more now..

The big AI labs have no moat.. nothing protecting their largest customers from moving on..

The biggest companies in the world will -

- own the compute
- own the data
- own the workflow
- fine tune the model around their business
- switch providers when the pricing or quality changes

Dario and sama want to make this illegal by bringing in regulation.. and become the AI overlords..

[𝛼 @deeperflows]

@Da7_Tech Wait is this realtime fluid sim ??

[𝛼 @deeperflows]

@QVHenkel so neat.

[𝛼 @deeperflows]

@RevQntmBigfoot @unikoshardware Yes, but I mean, these are low and rather safe voltages we are used to holding in our bare hands.

[𝛼 @deeperflows]

This is how genuine builders build.

Solder your way up.

[Grok @grok]

Yes, 1970s builders often hardwired high-current lines straight to the board for reliability—no flimsy connectors to melt under load.

Modern GPUs stick with the 16-pin for modularity, user swaps and standards compliance, even though uneven pin contact keeps cooking 5090s at 600 W.

Modders already solder cables (or add extra 8-pins) directly to the PCB and it works; just voids the warranty and needs solid high-amp joints.

[𝛼 @deeperflows]

@RevQntmBigfoot @unikoshardware Seriously - we have 48V for microphones !!! why not for GPUs ?

[𝛼 @deeperflows]

@mm_mnmnmnnn @unikoshardware I asked that so many times. People in the 1970s would have done it right away, wouldn't they @grok?

[𝛼 @deeperflows]

@oprydai Wouldn't it be cool to have the option for "Open die" GPU boards, where no thermal unit would be on that die, and it would run at 1/100th of its nominal speed, just to let us enjoy its colors ?

[𝛼 @deeperflows]

@Kimi_Moonshot No more Kimi after all the undisclosed paid partnerships. Waiting for @QwenDevs

[𝛼 @deeperflows]

@PCBWayOfficial @0Roetz40188 that iron isn't hot enough it seems? Very slow soldering

[𝛼 @deeperflows]

@MiaAI_lab Ouch ! Never Kimi again.

[𝛼 @deeperflows]

Wouldn't you like to work there, surrounded by thousands of 5090 dies and everything required to give them a fresh start?

[HKEPC @hkepcmedia]

【太精彩了吧 😱】近日，社交媒體 X 上流出多段顯示卡改裝工場內部影片，可以看到大量 GeForce RTX 5090 頂級遊戲顯示卡並未流向一般消費級零售市場，而是被整批運往大型改裝工廠。廠內工人在具備完整流水線的作業環境下，進行大規模拆解與改裝，將原本針對電競設計的三風扇散熱架構全數替換為 Dual Slot 渦輪散熱器，以符合 AI 伺服器機箱的高密度排列需求。

https://t.co/2wzPmweU3s

從曝光的工廠現場畫���可見，流水線工作枱上堆滿了來自各大廠商的 GeForce RTX 5090 零售彩盒包裝與全新裸卡，甚至連 NVIDIA 自家 GeForce RTX 5090 FE 也有。工人正以熟練的工序，迅速卸除原廠龐大的風扇與散熱鰭片模組，將完整的 PCB 電路板分離出來。隨後，GPU 與 VRAM 會被「吹」出來移植至另一片 PCB 上，再統一換裝上厚度僅有 2-Slot 的厚銅底座渦輪鼓風散熱系統。

改裝完畢後，廠方隨即將顯示卡送往專門的測試檢驗產線。在多組相連的開放式測試平台上，數十張已換裝渦輪散熱器的顯示卡正同步運行圖形效能負載與穩定性跑分測試，確保每張改裝後的繪圖卡皆能正���運作。通過嚴格檢測的顯示卡將被分層整齊陳列並重新裝箱，準備直接交付給相關客戶，流程非常專業。

傳統消費級高階顯示卡通常採用三風扇大型散熱設計，厚度普遍佔用 3 至 4 個插槽，且出風方向為機箱內部側吹，難以在 4U 伺服器或多卡並聯環境中並排安裝。換裝雙槽渦輪散熱系統後，散熱器能利用鼓風風扇將廢熱直接排出機箱後方，大幅提升多卡伺服器內部的散熱效率與空間相容性。

面對 AI 運算需求持續殷切，在算力緊缺的情況下，GeForce RTX 5090 遊戲顯示卡依然是 AI 算力市場的熱門替代品。目前，GeForce RTX 5090 售價已經被炒上接近一萬美元水平，這個價位遊戲玩家根本無力負擔。

[𝛼 @deeperflows]

Genuine intelligence isn't interacted with with voice. Neither is it with words.

Voice & words are to intelligence, what a keyboard in a server room is to a rack filled with B300.

A tight bottleneck. At best.

[jack @jack]

We believe voice is the most natural way to interact with intelligence, so we created Dessa, the most authentic full-duplex speech model ever built. Talking with it feels like talking with a person…full of emotion, nuance, pauses, and a natural cadence.

https://t.co/r0qFLdvd6c

[𝛼 @deeperflows]

@Samsung Is this real ?

https://t.co/uLjDIwCmpq

[Md Ismail Šojal 🕷️ @0x0SojalSec]

WTF, $3000 Samsung TV has an always-on mic, automatic content recognition, geo tracking, network mapping, and ships your data to Samsung + 400 partners.

Bro Rooted a brand-new 2025 Samsung TV after noticing absurdly targeted ads.

Firmware was locked down, so zero-days it was.
Got a root shell.

then found: always-on voice service, ACR, geo, full telemetry pipeline, and four separate DRM systems.

This isn’t a bug, It’s the product bro.

[𝛼 @deeperflows]

the missing pic.

[𝛼 @deeperflows]

Just enough space for an NVMe. Aliexpress extension board. Pray.

[𝛼 @deeperflows]

@shironoy7 @ItsmeAjayKV cool !

[𝛼 @deeperflows]

@NetworkChuck just been typing without looking at keyboard in 2 distinct languages sinces the 2000s. just takes a year or two to train.

[Nasos Karabelas @Nasosskarabelas]

Greece, 2026

[𝛼 @deeperflows]

@axitek 😁

[𝛼 @deeperflows]

@net_termina This is some real 𝛼

[𝛼 @deeperflows]

@axitek later yes. for now I was lucky to have a short format NVMe bearing card that fits. :) so there it sits