TechTuber Builds Colossal PC Case, Housing Human 'Figurine

TechTuber Soda Baka constructed a colossal PC tower large enough for a human to work and play inside; the TechTuber posed within, making it appear as if a figurine was inside the RGB-lit structure. The build was influenced by Baka’s little cousin being barked at by his mom, 'You play computer games all day, why don’t you just live in a computer!' or something similar.

Credit: Soda Baka on Bilibili

Shared on Bilibili, the elaborate project also suggested a strategic video sponsorship deal for air conditioners, coinciding with warmer months in the Northern Hemisphere. The human-habitable PC tower project began with Baka sketching and modeling before construction of huge components like case fans, GPUs, AiO CPU coolers, and RAM sticks.

Baka incorporated splashes of RGB lighting and installed real-PC elements, including a compact desk where he was seen playing a game on a screen attached to the AiO. To simulate PC-scale heat, hot-coal sauna installations were used. Baka boosted heat and humidity with a bottle of water, and apparently, it didn’t take long for the enclosed room’s temperature to rise above 38 degrees Celsius.

With the colossal fans and coolers being non-functional, Baka activated an installed 12-kilowatt AC unit, which offered 820 cubic metres per hour air circulation to cool the "PC system." The atmosphere inside swiftly changed to become "like a spring evening in the North," of China, leading to the impression that the project might have been a promotional video for a Midea 3rd Gen Pro AC unit.

---

• TechTuber Soda Baka built a human-sized PC tower.

• The project allowed Baka to work and play inside the RGB-lit case.

• Internal temperatures rose above 38 degrees Celsius, requiring a 12-kilowatt AC unit.

Source: TOMS HARDWARE

Chinese AI Startup DeepSeek Seeks Billions in First Funding Round

DeepSeek, a Chinese artificial intelligence startup, could achieve a valuation of up to USD 50 billion in its initial fundraising effort. This move marks a significant shift from its long-standing strategy of declining external investment.

Credit: UNSPLASH

China's national artificial intelligence fund, established in January last year, is in discussions to be a lead investor in DeepSeek's fundraising. The fund was founded with 60 billion yuan (USD 8.8 billion).

The startup aims to raise USD 3 billion to USD 4 billion from the funding round. These funds are intended to fuel its computing capabilities and enhance employee benefits.

Chinese tech giant Tencent Holdings has also been in talks to invest in DeepSeek. All sources involved in these discussions declined to be named due to the confidential nature of the information.

The China Integrated Circuit Industry Investment Fund, the primary backer of the national AI fund, declined to comment. Tencent also declined to comment.

The Financial Times previously reported that the China Integrated Circuit Industry Investment Fund was in talks to lead DeepSeek's first fundraising. This report suggested a potential valuation of about USD 45 billion for the frontier AI laboratory.

This maiden fundraising round occurs as DeepSeek faces increasing competition from domestic rivals. These competitors include tech giants such as ByteDance and Alibaba, alongside emerging startups like MiniMax and Moonshot AI.

These rivals have secured billions from either private or public markets, allowing them deep pockets.

Founder Liang Wenfeng has historically preferred to fund and operate DeepSeek as a research laboratory. Its budget previously came from his quant hedge fund, High-Flyer, rather than a Chinese tech conglomerate or an initial public offering.

Wenfeng is directly involved in the current fundraising talks.

Some of DeepSeek's researchers have been poached by competitors. For example, Luo Fuli departed the startup last year to lead Xiaomi's artificial intelligence model team MiMo.

The fast-moving AI industry has largely moved beyond and recreated the low-cost, highly efficient open-source chatbot models that contributed to DeepSeek's global breakthrough early last year.

The industry's focus has now shifted to agents capable of more complex tasks with less human intervention. These agents require substantially more computing power to operate.

DeepSeek claimed its next-generation, agent-suited V4 model "redefined the state-of-the-art" for open-source models last month. However, third-party evaluations indicated it lagged behind leading models from some United States and Chinese competitors.

Notably, the release of V4 did not trigger the global tech share selloff observed last year after its predecessors, V3 and R1.

---

• DeepSeek, a Chinese artificial intelligence startup, is seeking its first external funding, with a potential valuation of up to USD 50 billion.

• This fundraising effort marks a significant shift from DeepSeek's long-standing strategy of declining outside investment, previously relying on founder Liang Wenfeng's quant hedge fund.

• China's national artificial intelligence fund and Chinese tech giant Tencent Holdings are in discussions to invest, potentially contributing USD 3 billion to USD 4 billion.

Source: REUTERS

Anthropic Secures SpaceX Data Centre Deal for AI Development

Anthropic has reached an agreement to utilise computing resources from Elon Musk’s SpaceX, marking a détente with its former critic. This deal provides a significant boost for both companies in the highly competitive artificial intelligence race.

Credit: SPACEX

Under the agreement, Anthropic will access the full computing power of SpaceX’s Colossus 1 facility in Memphis, Tennessee. This facility contains more than 220,000 Nvidia processors.

The arrangement will provide Anthropic, the maker of the Claude chatbot, with 300 megawatts of new computing capacity within a month. This helps Anthropic alleviate capacity constraints following a surge in demand for products such as its AI coding tool, Claude Code.

The deal also provides IPO-bound SpaceX with a prominent customer, supporting its artificial intelligence ambitions. Both companies expressed interest in collaborating on developing multiple gigawatts of space-based orbital data centres in the future.

Ryan Mallory, chief executive officer of data centre operator Flexential, commented on the significance of this ambition. Mr. Mallory stated, "The fact that serious companies are even discussing compute capacity in space tells you how aggressively the market is searching for power and scale."

This future endeavour is a key goal for Mr. Musk and a major driver for SpaceX’s initial public offering. However, it is expected to be highly capital intensive and technically challenging.

Elon Musk stated on X that he decided to lease computing power after spending time with Anthropic leaders. He was impressed by their work to ensure Anthropic’s Claude AI is "good for humanity."

Mr. Musk wrote, "No one set off my evil detector," a reference to his previous concerns about AI systems. He is currently involved in a court dispute with OpenAI and its Chief Executive Officer Sam Altman regarding its mission to develop AI for the benefit of humanity.

He added that SpaceX has moved its AI training efforts to Colossus 2. Mr. Musk indicated he would offer computing capacity to other AI companies committed to benefiting humanity, akin to how SpaceX launches satellites for competitors with "fair terms and pricing."

These comments represent a significant shift from remarks Mr. Musk made in Feb., when he accused Anthropic’s AI of bias. At that time, he posted on X, "Frankly, I don’t think there is anything you can do to escape the inevitable irony of Anthropic ending up being misanthropic."

Anthropic unveiled a new Claude AI feature called "dreaming" at a developer day in San Francisco. This feature helps its AI systems learn by reviewing work between sessions, spotting patterns, and updating files that store user preferences and other context.

The "dreaming" feature is available as a research preview and comes with Anthropic’s software for managing agents. Agents are AI programmes designed to perform tasks with minimal human involvement.

Buoyed by the extra capacity from the SpaceX deal and other recent agreements, Anthropic is doubling Claude Code’s rate limits for its paid plans. The company is also removing peak-hour usage caps for Pro and Max accounts.

Additionally, Anthropic is sharply increasing the volume of requests developers can make to its Claude Opus models. This push is part of Anthropic’s strategy to secure more business customers.

The popularity surge for Claude Code has intensified competition in the market. This has prompted OpenAI to scale back some efforts, such as its Sora video-generation tool, to focus more on the fast-growing market for AI-powered coding.

Boris Cherny, Anthropic’s head of Claude Code, demonstrated how developers could set up "routines" to schedule Anthropic’s AI computer programmer for action. Mr. Cherny demonstrated this on stage at the San Francisco event.

He commented, "The default isn't, 'I'm going to prompt Claude Code.' The default is now, 'I will have Claude prompt Claude Code.'" Mr. Cherny concluded his keynote address with a message on AI’s productive power.

He stated, "The capability is already here. The gap left is how fast we put it to work."

---

• Anthropic secured a deal to use computing resources from SpaceX’s Colossus 1 facility.

• This agreement provides Anthropic with 300 megawatts of new capacity and offers SpaceX a significant customer for its AI ambitions.

• Anthropic announced the "dreaming" AI feature and is expanding Claude Code’s capabilities.

Source: REUTERS