Google develops AI chips as Nvidia expands chip supply

Key Takeaways

• Google is partnering with Marvell Technology to develop two new AI chips, including a memory processing unit and an advanced TPU, to diversify its supply chain and reduce reliance on Nvidia and Broadcom.
• Google DeepMind's Gemini models offer a 'native advantage' for multimodal applications, with research focusing on areas like agentic worlds and creative technologies.
• Google co-founder Larry Page's 2007 prediction that AI progress hinges on massive computation, not just algorithms, is validated by Google's custom Tensor Processing Units (TPUs) developed since 2015.
• The 'No Fakes' bill has gained enough support to advance out of committee, indicating growing momentum for comprehensive AI regulation.
• A Federal Reserve study suggests AI is reshaping job tasks and wages, particularly in white-collar sectors, by augmenting roles rather than causing widespread layoffs.
• PayPal co-founder Max Levchin predicts software engineers will become 'software sculptors,' guiding AI in code generation with human taste and understanding.
• Marketing teams often face an 'execution gap' with AI, struggling to achieve desired results because they add AI to existing workflows instead of redesigning them for human-AI collaboration.
• Economist Alex Imas is optimistic about AI's impact on jobs, theorizing that cheaper goods and services will shift spending and employment towards human-centric 'relational sector' services.
• Altamonte Springs is using AI to save taxpayer money, with its Police Department achieving a 14% ROI through AI-powered report writing and an AI bot (Herman Resources) saving HR staff time.
• Tattoo artists are incorporating AI as a creative tool for generating concepts and feedback, enhancing their craft by combining multiple AI-generated references for unique designs.

Google explores Marvell partnership for new AI chips

Google is reportedly in talks with Marvell Technology to create new chips specifically for running AI models. This move aims to diversify Google's chip supply chain, which currently relies heavily on Broadcom. The new chips would complement Google's existing Tensor Processing Units (TPUs), like the recently announced Ironwood, by potentially targeting different AI workloads or cost points, especially for inference tasks. This strategy allows Google to build a more robust and flexible supply chain for its growing AI needs.

Google in talks with Marvell for new AI inference chips

Alphabet's Google is reportedly in discussions with Marvell Technology to develop two new chips for AI inference. One chip would be a memory processing unit to work with Google's existing TPUs, while the other would be a new TPU designed for running AI models. These talks come as Google seeks to reduce its reliance on Nvidia and diversify its AI chip partnerships. The companies aim to finalize designs for the memory processing unit soon.

Google and Marvell discuss new AI chips for efficiency

Google is in talks with Marvell Technology to create two new chips aimed at improving AI model efficiency. One proposed chip is a memory processing unit to assist Google's Tensor Processing Units (TPUs), and the other is a more advanced TPU. This potential partnership is part of Google's broader effort to reduce dependence on Nvidia and explore collaborations with other chipmakers. The new chips would focus on inference, the process of running trained AI models, which is becoming increasingly critical as AI applications expand.

Google DeepMind VP shares AI's future intelligence vision

Raia Hadsell, VP of Research at Google DeepMind, discussed the future of artificial intelligence and intelligence itself at AI Engineer Europe. She highlighted Google DeepMind's Gemini models and their 'native advantage' for multimodal applications. Hadsell also emphasized the role of simulation in AI research, referencing past successes with games and recent work on weather forecasting with GenCast. DeepMind's research spans areas like agentic worlds, AI for humans, sustainability, creative technologies, and advanced models.

Larry Page's 2007 prediction on AI computation proves true

In 2007, Google co-founder Larry Page predicted that artificial intelligence progress would rely heavily on massive computation rather than just clever algorithms. He compared AI's complexity to DNA, suggesting the core algorithms are not overly complex but require significant processing power. Google's development of custom Tensor Processing Units (TPUs) since 2015 aligns with this prediction, providing specialized hardware for AI workloads. This foresight highlights computation as a key engineering challenge for achieving AI advancements.

AI regulation bill 'No Fakes' gains momentum

A legislative bill known as 'No Fakes' is gaining support as a potential foundation for comprehensive artificial intelligence regulation. The bill has reportedly secured enough votes to advance out of committee. This development signals growing momentum behind efforts to establish a regulatory framework for AI technologies.

Marketing teams face AI execution gap

Many marketing teams struggle to achieve desired results from AI despite increasing budgets and pilot programs. The main issue is an 'execution gap,' where organizations try to add AI to existing workflows instead of redesigning them. High-performing teams treat AI as a collaborative tool, not a replacement, and focus on redesigning workflows, fostering human-AI collaboration, and building governance early. Successful AI integration requires rethinking how work is done, how decisions are made, and how teams align around outcomes.

AI reshaping jobs but not causing mass layoffs yet, Fed study finds

A recent Federal Reserve study suggests that artificial intelligence is currently reshaping job tasks, wages, and hiring practices rather than causing widespread layoffs. The study indicates that AI is more likely to augment jobs, particularly in white-collar sectors, by changing how work is performed. While long-term effects remain uncertain, the immediate impact focuses on productivity gains and potential shifts in wage structures, with lower-wage service jobs being less affected in the short term.

Software engineers to become 'sculptors' in AI era, says Max Levchin

PayPal co-founder Max Levchin believes software engineers will evolve into 'software sculptors' in the age of AI. He argues that while AI can generate code, the human element of taste, elegance, and understanding is crucial for guiding the process. Levchin suggests that future engineers will focus less on syntax and more on conversing with AI to shape desired outcomes. This role requires a strong foundation in computer science to ensure the creation of elegant and effective software.

Economist finds hope for AI's impact on jobs

University of Chicago economist Alex Imas, initially concerned about AI's effect on labor markets, now sees a reason for optimism. He theorizes that as AI makes goods and services cheaper, spending and employment will shift towards human-centric services that AI cannot easily replicate. Citing examples like Starbucks' focus on human interaction, Imas suggests that demand will grow for services with a distinct human element. This migration towards the 'relational sector' could lead to a positive economic future despite AI advancements.

Altamonte Springs uses AI to save taxpayer money

Altamonte Springs is implementing artificial intelligence across various city departments to improve efficiency and save taxpayer money. The city's Police Department is using AI for faster, more accurate report writing and multilingual communication, resulting in a 14% return on investment. Additionally, an AI bot named Herman Resources assists city employees with HR inquiries, saving significant staff time. Altamonte Springs is also pioneering AI-powered site plan reviews, reducing review times from weeks to minutes.

AI influences tattoo culture and artistry

Tattoo artists are increasingly incorporating artificial intelligence into their creative process, discussing its growing role at the Black Web Ink Tour in Fargo. While AI is used for feedback and generating initial concepts, artists emphasize that it serves as a tool to enhance their craft rather than replace artistic skill. The tattoo industry is experiencing rapid growth, with AI raising questions about its future impact. Artists often combine multiple AI-generated references to create unique designs, highlighting the blend of technology and traditional artistry.

Sources

• Google is in talks with Marvell to build custom AI inference chips as it diversifies beyond Broadcom
• Google in talks with Marvell to build new AI chips for inference, The Information reports
• Google in Talks With Marvell to Build New AI Chips for Inference
• Google DeepMind VP on AI's Future of Intelligence
• How Larry Page Had Correctly Predicted In 2007 That Increased Computation — Instead Of Clever Algorithms — Would Lead To AI Progress
• Tech: ‘No Fakes’ gains ground
• The Execution Gap Holding Back AI In Modern Marketing Teams
• AI won't trigger mass layoffs yet, Fed study says
• Software Engineers Will End Up Becoming ‘Software Sculptors’ In Age Of AI: PayPal Co-founder Max Levchin
• The economist who was terrified of AI just found a rare reason for hope
• Altamonte Springs embraces AI technology to save taxpayer money, City Manager says
• Beyond the Needle: How A.I Is Reshaping Tattoo Culture