The artificial intelligence sector is experiencing rapid expansion and transformation, marked by significant hardware advancements and strategic partnerships. NVIDIA is pushing the boundaries of AI infrastructure with its Vera Rubin NVL144 MGX servers, designed for gigawatt AI factories and utilizing NVIDIA Kyber to connect 576 Rubin Ultra GPUs. These facilities will employ 800-volt direct current (VDC) power systems, a technology also supported by Navitas Semiconductor's new power devices, to enhance energy efficiency and scalability, drawing parallels to its use in the electric vehicle and solar industries. In parallel, OpenAI has secured a multiyear deal with Broadcom for custom AI chips and networking equipment, aiming to add 10 gigawatts of AI data center capacity by the second half of 2026. This move by OpenAI, alongside other AI companies, highlights a trend of securing specialized hardware to advance AI capabilities. Meanwhile, the AI market's rapid growth has led former Intel CEO Pat Gelsinger to describe it as a bubble, though he anticipates it will persist for several years, noting that businesses have yet to see substantial benefits. The impact of AI on the workforce is also a key discussion point, with current data suggesting AI is more likely to change job tasks than eliminate roles, though the long-term effects are still unfolding. A new challenge emerging is 'workslop,' low-quality AI-generated content that creates extra work and frustration for employees. In the creator economy, tools from Meta and OpenAI are splitting creators into those who embrace AI for scale and those who emphasize human elements. Cybersecurity firms like DeepTempo and Cribl are partnering to combat AI-driven threats, enhancing detection of sophisticated AI attacks. Businesses are being urged to adopt AI promptly to avoid operational inefficiencies and loss of market share, with pilot projects recommended for demonstrating AI's value.
Key Takeaways
- NVIDIA is introducing Vera Rubin NVL144 MGX servers for gigawatt AI factories, connecting 576 Rubin Ultra GPUs.
- New AI data centers are adopting 800 VDC power systems for improved energy efficiency and scalability, with Navitas Semiconductor providing key power devices.
- OpenAI has partnered with Broadcom for custom AI chips and networking, aiming to add 10 gigawatts of AI data center capacity by late 2026.
- Former Intel CEO Pat Gelsinger views the current AI market as a bubble that will last several more years.
- AI is currently changing job tasks rather than eliminating entire roles, though its long-term employment impact is still being assessed.
- 'Workslop,' or low-quality AI-generated content, is creating extra work and frustration for employees.
- The creator economy is seeing a division, with some creators leveraging AI tools like those from Meta and OpenAI, while others focus on human elements.
- Cybersecurity startups DeepTempo and Cribl are collaborating to improve the detection of advanced AI-driven threats.
- Businesses risk falling behind by delaying AI adoption, facing increased costs and potential loss of market share.
- AI product development is evolving through five stages, from static ML to AI-first designs.
NVIDIA unveils Vera Rubin servers for gigawatt AI factories
NVIDIA is introducing the Vera Rubin NVL144 MGX servers, designed for efficient gigawatt AI factories. These servers, supported by over 50 partners, will use NVIDIA Kyber to connect 576 Rubin Ultra GPUs for increased AI demands. The new 800-volt direct current data centers will improve energy efficiency and scalability. NVIDIA plans to share its rack design as an open standard with the Open Compute Project.
800 VDC power system to boost AI factory efficiency
The rise of AI has created 'AI factories' that need more power, leading to a shift towards 800 Volts direct current (VDC) power systems. This new architecture, along with energy storage, aims to handle the high power demands and rapid load swings of AI workloads. Using 800 VDC reduces copper use and costs, improves energy efficiency by eliminating conversion steps, and creates a simpler, more reliable system. This approach is already used in the electric vehicle and solar industries.
Navitas Semiconductor powers NVIDIA's 800 VDC AI factories
Navitas Semiconductor has launched new power devices, including 100 V GaN FETs and 650 V GaN and SiC devices, to support NVIDIA's 800 VDC AI factory architecture. These components are designed for different stages of power conversion within the AI data centers. The 100 V GaN FETs are for GPU board DC-DC stages, while the higher voltage devices handle front-end conversion. Navitas is using a 200 mm GaN-on-Si process for scalable production.
OpenAI and Broadcom partner for AI chips and networking
OpenAI has signed a multiyear deal with Broadcom to develop custom chips and networking equipment. This partnership aims to add 10 gigawatts of AI data center capacity, with new server racks expected to be deployed in the second half of 2026. This collaboration focuses on advancing AI capabilities through specialized hardware.
OpenAI and Broadcom agree on chip deal
OpenAI has entered into a new agreement with chipmaker Broadcom. This deal is part of a broader trend of AI companies securing specialized hardware. The report also touches on tech stocks recovering and Warner Brothers Discovery rejecting a takeover offer.
AI product development evolves through five stages
Companies are progressing through five stages of AI maturity in their products: static ML, dynamic ML, multiple ML, continuous ML, and AI-first. Initially, AI models are pre-trained and updated infrequently. Later stages involve models that adapt to context, orchestrate multiple models for complex tasks, and continuously improve with new data. The final stage, AI-first, designs products around AI from the beginning, like autonomous vehicles.
AI is changing jobs, not eliminating them yet
While fears of AI taking jobs are widespread, current data shows AI is more likely to change job tasks than eliminate entire roles. It takes time for companies and workers to adopt new technologies, and significant labor market impacts from AI have not yet been observed. Some workers are already seeing AI augment their tasks, potentially freeing up time for higher-value work. Experts suggest it will take several more years to fully understand AI's long-term effect on employment.
AI 'workslop' wastes time and breeds resentment
A new term, 'workslop,' describes AI-generated content that lacks substance and creates extra work for employees. Researchers found that 40% of employees received workslop in the past month, leading them to question peers' intelligence and the value of AI. This often results in employees having to redo reports or hold meetings to clarify confusing AI-generated memos. While AI can benefit work, companies need to address the issue of low-quality AI content.
Former Intel CEO: AI market is a bubble for years
Former Intel CEO Pat Gelsinger believes the current AI market is a bubble, but he doesn't expect it to burst for several years. He noted that while there's been a significant industry shift towards AI, businesses haven't yet seen major benefits. Gelsinger, who stepped down as Intel CEO in late 2024, acknowledged that Intel was late in adopting AI. Other tech leaders have also expressed concerns about the AI market's valuation and sustainability.
DeepTempo and Cribl partner to fight AI threats
Cybersecurity startup DeepTempo and data management company Cribl have partnered to combat advanced AI-driven threats. Their integrated solution uses DeepTempo's LogLM model and Cribl's data collection tools to improve the detection of polymorphic and agentic AI threats. This collaboration aims to provide security teams with high-fidelity detections, faster investigations, and cost savings by streamlining data management and analysis.
AI splits creators into two distinct paths
The creator economy is dividing as AI tools like Meta's Vibes and OpenAI's Sora emerge. Some creators are embracing AI to increase production speed and scale, while others are focusing on human elements like vulnerability and imperfection. This split is forcing creators to decide whether to use AI as a tool to enhance their existing work or to create entirely new forms of content. The future of content creation may depend on the intent behind using these powerful AI technologies.
Businesses risk falling behind by delaying AI adoption
Businesses that delay adopting artificial intelligence face significant risks, including rising costs, loss of market share, and difficulty retaining talent. AI is becoming a necessity for competitiveness, and inaction leads to operational inefficiencies and missed revenue. Pilot projects offer a low-risk way to demonstrate AI's value, while structured planning with readiness assessments and governance frameworks ensures successful implementation. Companies need to act now to avoid falling behind in the rapidly evolving AI landscape.
Sources
- NVIDIA, Partners Drive Next-Gen Efficient Gigawatt AI Factories in Buildup for Vera Rubin
- Building the 800 VDC Ecosystem for Efficient, Scalable AI Factories
- 800VDC Power: Navitas' 100V GaN FETs and 650V GaN/SiC Devices Enable NVIDIA AI Factories
- Bloomberg Intelligence: Open-AI, Broadcom Sign 10-Gigawatt Pact for Chips, Networking
- Bloomberg Tech: OpenAI, Broadcom Ink Chip Deal
- The AI-driven company: the AI-enabled product
- Don't panic: AI isn't here for your job (yet)
- How ‘workslop’ is wasting workers’ time and creating AI resentment
- Former Intel CEO says AI is a bubble that won't pop for 'several years'
- DeepTempo and Cribl team up to counter agentic AI threats with data-driven security
- The creator is splintering as AI forces a new reckoning
- AI is moving fast. Is your business falling behind?