Samsung Plans a Colossal 2.655 Trillion Won Investment in South Korea
Mark your calendars for June 29th—Samsung just dropped a massive bombshell: a 2.655 trillion KRW investment plan. The goal? To turn South Korea into the world’s ultimate high-tech manufacturing hub.Here’s the breakdown: they’re plowing 2.03 trillion KRW into the semiconductor cluster around Seoul (think Pyeongtaek and Yongin), while dumping another 625 trillion KRW into regions like Honam, Chungcheong, and Yeongnam.
Lee Jae-yong, Chairman of Samsung Electronics, spilled the beans that current capacity just isn’t cutting it for market demand. They’re eyeing a new facility in Gwangju, plus robotics in Gyeongsan, biotech in Incheon, battery tech in Ulsan, and semiconductor substrates in Busan. Basically, they’re going all-in everywhere.
On the same day, Kim Jin-gwan, Korea’s Minister of Trade, Industry and Energy, confirmed the government is chipping in with about 800 trillion KRW. The plan? Building four new semiconductor wafer fabs in the southwest region, with Samsung Electronics and SK Hynix each taking charge of two.
DeepSeek API Users Get Shocked: V4正式版 Peak Prices Double
June 29th brought some startling news from the developer community. According to reports from Lanjing News, several devs confirmed they received upgrade emails from DeepSeek. The big reveal? The official V4 version launches mid-July, bringing a whole new pricing strategy with “peak vs. valley” rates to manage resources better.
Here’s the kicker for those of you running APIs: while V4Pro and V4Flash keep their standard rates during off-hours, expect costs to skyrocket to 2x during peak times. That means paying double between 9 AM – 12 PM and 2 PM – 6 PM daily. Not exactly a budget-friendly surprise!
Musk Drops Bombshell: Grok 4.5 Enters Private Beta at SpaceX & Tesla, Claims to Beat Opus
Elon Musk took to X on June 28th local time to announce that his latest brainchild, Grok 4.5, has officially started internal testing at both SpaceX and Tesla. The specs are wild: built on a 1.5 trillion parameter V9 base model, it even ingested data from the hot new coding tool, Cursor, during its training.
The early verdict? Performance is already neck-and-neck with, or potentially surpassing, Anthropic’s flagship Claude Opus. While Reinforcement Learning (RL) is still actively sharpening the model’s edges, the “Grok Build” benchmark suite is maturing fast. Musk also teased that SpaceX will roll out a brand-new, fully trained model every single month for the rest of this year. Buckle up, because things are moving fast.
Hewlett Packard Partners with OpenAI: Full-Scale Frontier Integration to Accelerate Business Transformation
On June 28th, Hewlett Packard made a strategic move that could shake up the enterprise landscape. They’ve teamed up with OpenAI to deploy the OpenAI Frontier platform across their global operations. This isn’t just a small pilot; it’s a full-blown integration designed to supercharge HP’s transformation and growth roadmap.
Under this partnership, the Frontier platform will be woven into every facet of HP’s business, aiming to deliver a top-tier customer experience while streamlining internal workflows. It’s a clear signal that AI is no longer optional—it’s central to their future.
Zhipingfang Closes Latest Funding Round at Nearly 5 Billion RMB
Breaking news on June 29th: the embodied AI powerhouse Zhipingfang just secured a massive funding round totaling nearly 5 billion RMB (approx. $700 million). This capital influx comes from a who’s-who of heavy hitters, including the National SME Fund, China Cultural Industry Fund, Guangdong AI Fund, Nanshan Strategic New Industries Investment, and various Greater Bay Area funds.
Even more impressive? Veteran investors like DVCapital, Dunhong Asset, Daode Investment, and others doubled down. On the industrial side, giants like CP Group (Sinopharm), Pharmaron, Moutai Group, China Merchants Capital, and Wuzhou Xinchun joined as co-investors. Several key players in Tesla’s supply chain also threw in significant support. With this cash injection, Zhipingfang is set to accelerate the iteration and mass production of their “Robot Brain.”
Samsung Electro-Mechanics Eyes 500 Billion Won AI Server Order, Joins Forces with Sumitomo Chemical
Reports from June 29th indicate Samsung Electro-Mechanics is in the final stages of negotiating a roughly 500 billion KRW contract to supply MLCCs for AI servers to a major US tech giant. If signed, this deal alone would represent 10% of their components division’s revenue from last year.
Simultaneously, this week Samsung Electro-Mechanics is set to sign a joint venture agreement with Japan’s Sumitomo Chemical. Together, they’ll invest 500 billion KRW (with Samsung holding the majority stake at 300 billion KRW) to build a glass substrate plant in Pyeongtaek, South Korea. Production is scheduled to kick off by early next year.
Compute Crisis Hits Hard: Google Limits Meta’s Access to Gemini Amidst Shortage
The AI gold rush is hitting a snag. According to CCTV Finance and the UK’s Financial Times (report dated June 28), Google has started throttling Meta’s access to its powerful Gemini models. Why? Simply put, Meta’s compute demands have blown past what Google’s infrastructure can currently handle.
This supply crunch has thrown a wrench into several of Meta’s internal AI projects, forcing development timelines to slip. The situation lays bare a brutal truth: compute power remains the absolute bottleneck for the entire AI industry. Even with Google pouring money into infrastructure, they can’t keep up with the explosive demand. Data shows Google Cloud revenue hit $20 billion in Q1, but CEO Sundar Pichai admitted that the lack of compute is choking growth potential, with backlogged orders nearly doubling quarter-over-quarter.
