Big Shift in AI Stock Trade Drives Hunt for New Stars in Asia

Hewlett Packard Enterprise (HPE) has issued a disappointing sales outlook, projecting revenue between $9 billion and $9.4 billion and profit of 57 to 61 cents per share for the period ending in January, falling short of analysts' expectations of $9.88 billion in sales and 53 cents in profit. This news was highlighted by Patrick Moorhead on Bloomberg The Close.

Cloud-computing startup Fluidstack is reportedly in discussions to secure approximately $700 million in a funding round, which would elevate its valuation to around $7 billion, as per sources familiar with the matter.

Hewlett Packard Enterprise Co. shares fell in after-hours trading following a disappointing sales outlook for the current quarter, which did not meet high expectations for the AI server business. The company's forecast indicates a slowdown in server deals, raising concerns among investors.

OpenAI is facing intensified competition from Google, particularly with the rapid rise of Google's Gemini 3, which has gained 200 million users in just three months. In response, OpenAI CEO Sam Altman has declared a 'code red' for ChatGPT, emphasizing the urgent need for improvements to maintain its market position.

Leading artificial intelligence developers are increasingly focused on creating AI models that replicate human reasoning. However, recent research indicates that these advanced systems are significantly more energy-intensive, raising concerns about their impact on power grids.

ByteDance and DeepSeek, two prominent players in China's artificial intelligence sector, are pursuing markedly different strategies, highlighting the divergent paths within the industry. While ByteDance focuses on leveraging AI for content creation and user engagement, DeepSeek is emphasizing open-source AI models, such as its recent release that rivals GPT-5.

A new framework called Hypothesis-driven Backward Logical Reasoning (HBLR) has been proposed to enhance logical reasoning in large language models (LLMs) by integrating confidence-aware symbolic translation with backward reasoning. This approach aims to address inefficiencies in current forward reasoning paradigms, which often lead to redundant inferences and unreliable conclusions.

Recent research has highlighted the potential of Spiking Neural Networks (SNNs) to achieve adversarial robustness through natural gradient sparsity, revealing a trade-off between robustness and generalization in vision-related tasks. This finding suggests that under certain architectural configurations, SNNs can defend against adversarial attacks without explicit regularization.