June 13, 2025

news  
Read News
AMD AI Chips, Altman on AI, and AI Security

AMD is making significant strides in the AI market, introducing new AI chips and servers to compete with Nvidia. CEO Lisa Su anticipates the AI processor market to exceed $500 billion by 2028. OpenAI's CEO Sam Altman has expressed support for AMD's technology, stating that OpenAI will use AMD's MI300X and MI450 chips. AMD's new AI server, Helios, is set to launch in 2026 and will openly share networking standards with competitors like Intel. In addition to hardware advancements, AMD has launched the AMD Developer Cloud, providing access to its AI processors, and is collaborating with DigitalOcean to offer cloud GPUs. Meanwhile, the broader AI landscape is seeing both optimism and concern. Sam Altman predicts AI superintelligence is near and foresees major societal changes, while Nvidia CEO Jensen Huang disagrees with predictions of widespread AI-related job losses. There are also growing concerns about AI security, with reports indicating that many manufacturers lack adequate security controls. To address these risks, proposals have been made for a National Center for Advanced AI Reliability and Security. Other countries like Sweden are also boosting their AI infrastructure investments. Researchers are exploring ways to improve AI reliability and accuracy, such as FAU's method for cleaning bad data and TUM's approach of training AI with biological data from mice. Companies are also exploring how AI agents can boost performance in key business departments.

Key Takeaways

  • AMD is launching new AI chips (MI350 series) and servers (Helios) to compete with Nvidia, with the Helios server slated for 2026.
  • OpenAI will use AMD's MI300X and MI450 chips, according to CEO Sam Altman.
  • AMD CEO Lisa Su predicts the AI processor market will exceed $500 billion by 2028.
  • AMD has launched the AMD Developer Cloud and is partnering with DigitalOcean to provide cloud GPU access.
  • OpenAI CEO Sam Altman believes AI superintelligence is approaching, potentially leading to significant societal changes.
  • Nvidia CEO Jensen Huang disagrees with predictions of widespread AI-driven job losses.
  • A proposal suggests creating a National Center for Advanced AI Reliability and Security (CAISI+) to address AI risks.
  • Sweden is increasing its investment in AI infrastructure.
  • FAU has developed a method to improve AI reliability by cleaning faulty data.
  • Manufacturers are reported to lack adequate AI security controls, increasing risks like IP theft.

AMD's new AI server Helios challenges Nvidia with OpenAI support

AMD CEO Lisa Su introduced a new AI server called Helios for 2026, designed to compete with Nvidia. OpenAI's CEO Sam Altman said they will use AMD's MI300X and MI450 chips. Helios will openly share networking standards with competitors like Intel. AMD has acquired 25 AI-related companies and hired talent to boost its AI capabilities. Despite export curbs to China, AMD expects strong growth from AI chips.

AMD unveils AI server Helios as OpenAI uses its new chips

AMD CEO Lisa Su revealed a new AI server for 2026 to compete with Nvidia. OpenAI will use AMD's MI300X and MI450 chips. The new server, Helios, will have open networking standards shared with companies like Intel. Executives from xAI, Meta, and Oracle discussed using AMD processors. AMD acquired server builder ZT Systems and several AI startups to improve its AI technology.

AMD's next-gen AI chips get OpenAI's backing with Sam Altman

AMD revealed its next-generation AI chips, the Instinct MI400 series, launching next year. These chips can be assembled into a server rack called Helios. OpenAI CEO Sam Altman said his company will use the AMD chips. AMD's rack-scale setup will compete with Nvidia's Blackwell chips. AMD plans to offer its MI400 and MI355X chips at competitive prices.

AMD launches MI350 AI chips to challenge Nvidia, debuts cloud service

AMD launched its MI350 line of AI chips to compete with Nvidia's Blackwell processors. The company also introduced the AMD Developer Cloud, providing access to its AI processors. The MI350 line includes the MI350X and MI355X, offering improved AI compute performance. AMD's MI400 line of chips, set to launch in 2026, will compete with Nvidia's GB300 Blackwell Ultra processors.

AMD unveils AI server Helios as OpenAI taps its newest chips

AMD CEO Lisa Su introduced a new AI server for 2026 to compete with Nvidia. OpenAI will use AMD's MI300X and MI450 chips. The Helios server will openly share networking standards with companies like Intel. Executives from xAI, Meta, and Oracle discussed using AMD processors. AMD acquired server builder ZT Systems and AI startups to enhance its AI technology.

AMD CEO predicts AI processor market will top $500 billion by 2028

AMD CEO Lisa Su said the company's new MI350 AI processors can compete with Nvidia's chips. She expects the AI processor market to exceed $500 billion in the next three years. The MI355 chips, which began shipping this month, are much faster than previous versions. Su made these remarks at an event in San Jose, California.

OpenAI CEO Sam Altman says ChatGPT is more powerful than any human

OpenAI CEO Sam Altman believes humanity is close to building artificial super intelligence. He stated that ChatGPT is already more powerful than any human. Companies are investing billions in AI, creating a competitive market. Altman expects the 2030s to be very different, with robots potentially building other robots. He also anticipates job losses but believes people will adapt and new policies will emerge.

Sam Altman says AI Singularity is near, foresees major societal changes

OpenAI CEO Sam Altman predicts the arrival of artificial superintelligence is closer than expected. He believes society will gradually adapt to this change. Altman suggests the 2030s will be significantly different, with AI surpassing human intelligence. He acknowledges potential job losses but anticipates new job categories and increased wealth. Altman supports a universal basic income to help people adjust to this techno-utopian future.

Nvidia CEO disagrees with Anthropic's prediction of AI job losses

Nvidia CEO Jensen Huang disagrees with Anthropic CEO Dario Amodei's prediction that AI will eliminate many entry-level jobs. Huang believes AI will change jobs but create more opportunities by making companies more productive. He suggests that if AI is dangerous, its development should be open and monitored. Huang also believes AI can solve autonomous driving problems and envisions cars conversing with humans like ChatGPT.

How AI agents can boost performance in key business departments

Companies often fail to effectively use AI because they treat AI agents as tools instead of teammates. AI agents can handle many customer support questions, freeing up human teams. FuseBase CEO Pavel Sher recommends mapping workflows before implementing AI agents. AI agents can automate outreach in sales, provide 24/7 customer support, streamline onboarding, and improve customer success.

Proposal for a National Center for Advanced AI Reliability and Security

A proposal suggests creating a 'CAISI+' within the Department of Commerce to address AI risks. This center would require secure facilities, specialized talent, and a budget of $67-155 million annually. CAISI+ would evaluate AI models for catastrophic risks and provide emergency assessments to the President and NSC. It aims to drive AI reliability and security research to safeguard national interests.

Sweden boosts AI infrastructure after warnings of underinvestment

Sweden is increasing its investment in AI infrastructure after concerns about lagging behind other developed nations. Recent investments include a €10 billion AI datacentre and a £23 million procurement for a public AI factory. A consortium of Swedish corporations plans to build and run an AI supercomputer. AI Sweden emphasizes the need for public money to ensure small firms can access AI compute facilities.

FAU's CA-AI cleans bad data to make AI smarter and more reliable

Researchers at FAU developed a method to automatically detect and remove faulty labels from data before AI training. This technique uses a math approach to find unusual examples that don't fit. Removing these outliers ensures AI receives high-quality information. The method improves classification accuracy and can be used as a standard step in developing machine learning systems.

Report shows manufacturers lack AI security controls despite risks

A report reveals that few manufacturing organizations have implemented AI security controls. Only 17 percent have automated technical controls, and many rely on basic measures like training. Manufacturers underestimate sector-specific threats like IP theft. The report urges organizations to acknowledge reality, deploy verifiable controls, and prepare for regulation to address AI risks effectively.

AI learns from mice how to predict movements more accurately

Researchers at TUM discovered that artificial neural networks can predict movements better when trained with biological data from early visual system development in mice. Pre-training with retinal wave data improves a neural network's performance. Networks pre-trained with retinal waves predicted visual patterns more quickly and accurately. This approach enhances AI applications like autonomous driving and robotics.

DigitalOcean and AMD team up to boost AI with cloud GPUs

DigitalOcean and AMD are working together to give DigitalOcean customers access to AMD Instinct GPUs. Customers can use AMD MI300X GPUs now, and MI325X GPUs will be available later this year. These GPUs improve AI workloads. The collaboration aims to help digital native enterprises use AI in their applications with competitive pricing and simple setup.

Sources

AMD Nvidia OpenAI Lisa Su Sam Altman AI Chips AI Server Helios MI300X MI450 MI350 MI355X MI400 AI Processors AI Infrastructure AI Security AI Reliability AI Risks AI Agents Artificial Superintelligence ChatGPT Job Losses AI Datacenter AI Supercomputer AI Training Machine Learning Autonomous Driving Robotics Cloud GPUs DigitalOcean AI Workloads AI Development AI Compute AI Market AI Investment AI Models AI Applications AI Technology AI Startups AI Talent AI Performance AI Innovation AI Research AI Automation AI Solutions AI Capabilities AI Compute Performance AI Security Controls AI Predictions AI Integration AI Adoption AI Ecosystem AI Competition AI Collaboration AI Advancement AI Transformation AI Future AI Potential AI Impact AI Opportunities AI Challenges AI Strategy