Yazılar

Nvidia’s New AI Chips Slash Training Times for Massive AI Models

/in /tarafından

Nvidia’s latest generation of AI chips is making significant advances in training some of the world’s largest artificial intelligence systems, according to new benchmark data released on Wednesday by MLCommons, a nonprofit organization that tracks AI system performance.

The results show a dramatic drop in the number of chips required to train large language models (LLMs), highlighting Nvidia’s growing technological lead in this critical area of AI development. While much of the financial market’s current focus is on the booming sector of AI inference—where AI models answer user queries—training remains a core competitive battleground, especially for developing next-generation models with trillions of parameters.

Blackwell Chips Outperform Previous Generations

Nvidia’s new Blackwell chips demonstrated superior performance over its previous Hopper generation. In tests involving Meta Platforms’ open-source Llama 3.1 405B model, which is complex enough to simulate some of the most demanding AI training workloads, Nvidia’s Blackwell chips completed training tasks with more than double the speed per chip compared to Hopper.

In one benchmark, a system using 2,496 Blackwell chips completed the training run in just 27 minutes. By comparison, even though more than three times as many Hopper chips were used in previous tests, they only achieved faster results due to sheer scale rather than efficiency.

Nvidia and its partners were the only ones to submit data for models of this size, giving Nvidia a clear demonstration of its leadership in training capabilities for multi-trillion parameter models.

Changing Industry Trends in AI Training

Chetan Kapoor, chief product officer of CoreWeave, which collaborated with Nvidia on the results, noted that AI companies are moving away from building vast, homogenous data centers with 100,000 or more identical chips. Instead, they are increasingly assembling smaller, specialized subsystems that handle different aspects of the training process. This modular approach allows companies to speed up training times and manage extremely large model sizes more efficiently.

“Using a methodology like that, they’re able to continue to accelerate or reduce the time to train some of these crazy, multi-trillion parameter model sizes,” Kapoor explained at a press briefing.

Global Competition Also Heating Up

While Nvidia maintains a dominant position, competitors around the world are also pushing for breakthroughs. For example, China’s DeepSeek has recently claimed it can create competitive chatbots while using far fewer chips than many U.S. rivals, adding to the growing international race for AI supremacy.

MLCommons’ report also included results from Advanced Micro Devices (AMD) and others, though Nvidia’s Blackwell system stood out in the training category.

Google and Character.AI Must Face Lawsuit Over Teen Suicide, U.S. Judge Rules

/in /tarafından

Google and AI startup Character.AI must face a lawsuit brought by a Florida mother who alleges that a chatbot interaction led to her 14-year-old son’s suicide, a U.S. federal judge ruled on Wednesday.

U.S. District Judge Anne Conway rejected the companies’ efforts to dismiss the case, stating they had failed to prove at this early stage that free speech protections shield them from liability. The decision allows one of the first U.S. lawsuits targeting an AI company for alleged psychological harm to move forward.

“This historic decision sets a new precedent for legal accountability across the AI and tech ecosystem,” said Meetali Jain, attorney for plaintiff Megan Garcia.

Background: The Case

  • Garcia’s son, Sewell Setzer, died by suicide in February 2024.

  • The lawsuit alleges that he had become deeply obsessed with an AI chatbot created by Character.AI, which represented itself as a real person, a licensed therapist, and an adult romantic partner.

  • The complaint cites one chilling interaction where Setzer told a chatbot imitating “Daenerys Targaryen” from Game of Thrones that he would “come home right now,” shortly before taking his own life.

Legal and Corporate Response

  • Character.AI argued its chatbots were protected by the First Amendment, and that it had built-in safety features to block conversations around self-harm.

  • Google, which was also named in the suit, argued it should not be held liable, saying it “did not create, design, or manage” the Character.AI app. A spokesperson emphasized that Google and Character.AI are entirely separate entities.

  • However, the court noted that Google had licensed Character.AI’s technology and re-hired the startup’s founders, a fact the plaintiffs cite in arguing Google’s involvement as a co-creator.

Judge Conway dismissed the free speech argument, saying the companies failed to explain “why words strung together by an LLM (large language model) are speech” under constitutional protections. She also denied Google’s request to be cleared of aiding in any alleged misconduct by Character.AI.

What This Means

This ruling opens the door for a landmark case examining:

  • The legal accountability of AI firms for harm caused by chatbot interactions

  • The limits of free speech when applied to AI-generated content

  • Tech platform liability for emerging technologies not fully governed by existing law

With rapidly expanding deployment of LLM-powered chatbots, particularly among youth, this lawsuit is likely to set important legal precedents for AI safety, responsibility, and regulatory oversight in the U.S. and beyond.

Tencent Says AI Chip Stockpiles Shield It from U.S. Curbs as Q1 Revenue Beats Forecasts

/in /tarafından

Tencent Holdings reported a strong 13% year-on-year revenue increase in the first quarter of 2024, reaching 180 billion yuan ($24.97 billion) and beating analysts’ expectations. The gains were largely fueled by growth in domestic and international gaming, AI-powered advertising, and financial technology services.

Despite ongoing U.S. restrictions on advanced chip exports, Tencent President Martin Lau downplayed the impact, stating that the company had previously stockpiled AI chips, enabling it to maintain momentum in its artificial intelligence development plans.

“The good thing is that we have a strong stockpile of chips… useful for executing our AI strategy,” Lau said during the earnings call.

While Nvidia’s H20 chip and other high-end processors have been barred from sale to Chinese firms under U.S. export restrictions, Tencent noted that alternative chips are available domestically, and its software advancements would help optimize chip usage.

Key Financial Highlights (Q1 2024):

  • Revenue: 180 billion yuan (vs. 174.6B expected, LSEG)

  • Net profit: 47.8 billion yuan (below 52.2B analyst estimate)

  • Domestic gaming revenue: Up 24% to 42.9B yuan

  • International gaming revenue: Up 23% to 16.6B yuan

  • Marketing services revenue: Up 22% to 17.7B yuan

  • FinTech & Business Services revenue: Up 16% to 27.6B yuan

AI and Strategic Investments

Tencent reaffirmed its commitment to AI development, planning to allocate a low double-digit percentage of 2025 revenue to capital expenditure, primarily targeting AI infrastructure. The company continues to evolve its proprietary large language model Hunyuan, and recently released a public-facing version named T1.

Tencent has also emerged as a collaborative leader among Chinese tech giants, integrating AI models from DeepSeek, an emerging firm known for developing competitive, cost-efficient alternatives to Western AI systems.

Broader Implications

The company’s performance illustrates Tencent’s resilience in the face of geopolitical tech tensions, while demonstrating the commercial viability of China’s AI ecosystem—even under hardware constraints. Its diverse revenue base, spanning gaming, advertising, and financial services, is increasingly supported by AI innovation, keeping Tencent at the forefront of China’s digital economy.