Tencent Technology

Samsung Electronics Q3 2026 financial report: Profits hit a record high driven by memory chips, but the terminal business is under pressure; total shareholder return for the year is expected to be between 67.08 billion and 82.00 billion USD.

Alphabet's fiscal year 2026 Q3 revenue was $127.6 billion, but revenue, adjusted earnings per share, and free cash flow all fell short of Wall Street expectations, with capital expenditures continuing to increase, leading to a 9.79% drop in stock price after hours.

OpenAI released the mid-range model GPT-6.1 Terra: significant improvements in visual understanding and computer operation speed, with a single token price about one-tenth of GPT-6 Astra, but real-world reviews are polarized.

It can be said that what investors are buying is not Anthropic's profits today, but whether it can turn "the country of genius in the data center," a phrase from the prospectus, into cash flow in the corporate accounts under the triple constraints of grid bottlenecks, inference costs, and model reliability.

Tencent Technology has learned that DeepSeek's second round of financing is nearing completion but has not yet closed: the scale has been raised from the initial 50 billion yuan to nearly 100 billion yuan, with a pre-investment valuation maintained at around 500 billion yuan, and a few institutions are still vying for the last opportunity to enter. In addition, the company is expected to initiate its IPO as early as next year.

Meta's revenue in the third quarter of 2026 increased by 26% year-on-year to $64.6 billion, but large-scale investments in AI infrastructure caused free cash flow to drop from $10.6 billion in the same period last year to -$2.1 billion. The article breaks down why Zuckerberg would rather burn free cash flow to continue stacking computing power, as well as the logic behind this investment's returns.

Meta released Muse Spark Open: a mixed expert model with a total of 816 billion parameters and 42 billion active parameters, open weights, native support for multimodal, and a context window of up to 1 million tokens, known as the most powerful open-source weight model in the United States to date.

The South Korean stock market has significantly corrected, and the AI industry chain is facing multiple pressures such as the memory cycle peaking, poor revenue expectations for OpenAI, high capital expenditures, and retail investor margin calls.

In stark contrast to Anthropic's verbal emphasis on safety while aggressively releasing new versions, OpenAI was forced to hit the brakes on the same day.

What kind of security boundaries does AI development actually need?