Meta LLaMA Advances with New Chips While Google and Nvidia Power Robotics

The artificial intelligence landscape is experiencing a significant shift, with investor focus moving from companies that enable AI, like chipmakers, to those that adopt it across various sectors. Goldman Sachs predicts AI will boost S&P 500 earnings growth by 0.4% in 2026 and 1.5% in 2027, despite only 30-40% of large firms currently using AI. Sectors such as industrials, materials, consumer discretionary, big banks, and pharmaceuticals are expected to benefit most. However, economist Jason Furman voices concerns about a potential AI financial valuation bubble, questioning if scaling laws will consistently translate into economic gains, even with substantial spending on data centers and energy. He does not anticipate widespread job loss from AI adoption. Hardware advancements are crucial for this evolution. Engineers, including Subhasish Mitra and Tathagata Srimani, developed a new monolithic 3D chip built by SkyWater Technology. This chip is four times faster than comparable 2D chips in tests, with simulations suggesting up to twelve-fold gains for future versions. This innovation could lead to 100 to 1,000 times better energy efficiency and speed, vital for future AI systems like Meta's LLaMA. In robotics, Agility Robotics' Digit humanoid, standing 5 feet 9 inches tall and capable of lifting 35-pound boxes, uses AI for autonomous warehouse tasks and is currently deployed by Amazon and Mercado Libre. Agility Robotics also integrates AI tools from Google and Nvidia into its daily operations. Despite the US-China tech rivalry, American companies are increasingly adopting powerful, low-cost Chinese open-source AI models such as Alibaba's Qwen and DeepSeek's R1. Usage among US developers grew from 1.2% to nearly 30% by late 2025, with one business saving $400,000 annually by switching to Qwen. Even major players like Nvidia, Perplexity, and Stanford researchers have integrated Qwen into their projects. In sales, Jason Lemkin highlights that core strategies remain effective, but companies need updated approaches. He notes that AI sales tools became truly effective only after Claude 4 and GPT-4 emerged in March 2025, emphasizing the need for salespeople to be product experts. For retail traders, a panel at FMLS:25 found most are not yet using AI, as it cannot manage emotional losses, though tools like Fiscal.ai can process financial data, such as Nvidia's earnings, in minutes. On a personal level, one individual found ChatGPT helpful in managing negative thought spirals by introducing relational frame theory and a unique

Key Takeaways

• AI adoption will significantly influence business success and stock market performance in 2026, shifting investor focus from AI enablers to adopters.
• Goldman Sachs projects AI will boost S&P 500 earnings growth by 0.4% in 2026 and 1.5% in 2027, with only 30-40% of large firms currently utilizing AI.
• Economist Jason Furman expresses concern about an AI financial valuation bubble, questioning the translation of scaling laws into economic gains despite substantial spending on data centers.
• Agility Robotics' Digit humanoid robot, standing 5 feet 9 inches tall, uses AI for autonomous warehouse tasks, currently employed by Amazon and Mercado Libre, and leverages AI tools from Google and Nvidia.
• New monolithic 3D chips, developed by engineers including Subhasish Mitra and Tathagata Srimani, are 4x faster than 2D chips and could offer 100-1,000x energy efficiency for future AI systems like Meta's LLaMA.
• US firms are increasingly adopting low-cost Chinese open-source AI models like Alibaba's Qwen and DeepSeek's R1, with usage among US developers reaching nearly 30% by late 2025, leading to significant cost savings.
• Nvidia, Perplexity, and Stanford researchers have integrated Alibaba's Qwen into their projects, demonstrating the cross-border adoption of Chinese AI models.
• AI sales tools became effective after Claude 4 and GPT-4 in March 2025, requiring salespeople to be product experts and offer huge value before payment.
• While most retail traders are not yet using AI, tools like Fiscal.ai can process financial data, such as Nvidia's earnings, in minutes, speeding up access to information.
• The AI job market is expanding with accessible roles like AI Prompt Engineer (209% increase in UK postings) and AI Training Data Specialist, offering new career opportunities with low entry barriers.

Jason Lemkin says old sales strategies still work

Jason Lemkin spoke at SaaStr AI London, explaining that the core go-to-market playbook is not broken, but companies are using outdated 2021 versions. Successful AI companies like Vercel and Replit use experienced B2B sales leaders. AI sales tools only became effective after Claude 4 and GPT-4 in March 2025. Companies must offer huge value before payment and understand that competitive advantages now last months, not years. Salespeople need to be product experts because AI will replace those who are not.

AI helps stop negative thoughts and mental spirals

The author struggled with constant negative thought spirals since childhood, finding traditional therapy helpful but also discovering a new approach with ChatGPT. The AI introduced the author to relational frame theory, which explains how learned word relationships create painful thought patterns. Using ChatGPT, the author explored these loops and found a unique trick: narrating thought spirals like David Attenborough. This method helped create distance from burdensome thoughts and offered a new way to manage them. The author also worried about losing personal traits while seeking self-improvement, calling it the mindful zombie tradeoff.

Economist Jason Furman worries about AI financial bubble

Economist Jason Furman shared his concerns about the AI market in a Bloomberg Q&A, stating he is more worried about a financial valuation bubble than a technology bubble. Furman explained that high valuations need both great technology and profit potential. He questioned if scaling laws will continue to translate into economic gains, noting that huge spending on data centers and energy is real activity. Furman worries if AI will actually boost productivity, as it currently acts mostly as a demand side customer for the economy. He does not believe AI will cause widespread job loss, citing historical patterns.

Retail traders barely use AI says FMLS:25 panel

A panel at the Finance Magnates London Summit 2025 discussed how AI affects broker growth and retail traders. Experts like Kieran Duff noted that most retail traders are still learning basics and do not use AI, as it cannot manage emotional trading losses. However, Dor Eligula from BridgeWise showed that AI tools can increase user education and engagement, leading to lower churn. Roy Michaeli of WNSTN AI emphasized personalized AI support over chatbots, acting as a bionic arm for traders. Braden Dennis from Fiscal.ai highlighted how AI speeds up access to financial data, processing Nvidia's earnings in minutes. The panel concluded that while AI tools are advancing rapidly, retail traders are only beginning to explore their potential.

Agility Robotics uses AI for Digit humanoid and daily tasks

Agility Robotics uses artificial intelligence for its humanoid robot, Digit, which stands 5 feet 9 inches tall. Digit works in warehouses, moving and stacking boxes up to 35 pounds, and is currently used by Amazon and Mercado Libre. Pras Velagapudi, Agility Robotics' CTO, states that AI helps Digit observe, interpret, and respond to its surroundings without human help. While Digit is for industrial use now, the company aims for it to handle manual labor people do not want to do. AI and robotics have always been connected, with modern AI crucial for Digit's controls, balance, and object perception. Agility Robotics also uses AI tools from Google and Nvidia in its daily operations.

Engineers create 3D chips to boost AI speed

A team of engineers, led by Indian American professors Subhasish Mitra and Tathagata Srimani, developed a new monolithic 3D chip. This chip, built in a U.S. foundry by SkyWater Technology, features dense vertical wiring and stacked memory and computing units. It helps overcome the memory wall and miniaturization wall that limit traditional 2D chips. Hardware tests show the 3D chip is four times faster than comparable 2D chips, with simulations predicting up to twelve-fold gains for future versions. This breakthrough could lead to 100 to 1,000 times better energy efficiency and speed, crucial for future AI systems like Meta's LLaMA. The project involved collaboration from Stanford, Carnegie Mellon, University of Pennsylvania, and MIT.

US firms use Chinese AI despite US-China tech rivalry

Despite the US and China competing in AI, American companies are increasingly using powerful, low-cost Chinese open-source AI models. Usage of these models, like Alibaba's Qwen and DeepSeek's R1, grew from 1.2% to nearly 30% among US developers by late 2025. These models offer significant cost savings; one US business saved $400,000 annually by switching to Qwen. Even major players like Nvidia, Perplexity, and Stanford researchers have integrated Qwen into their projects. While the US government promotes American open models, US firms have largely moved away from open-source investment. Concerns about Chinese origins exist, but experts suggest open-source transparency helps build trust and data security is not a major issue.

AI adoption will shape business success in 2026

In 2026, AI adoption will heavily influence which companies succeed in the stock market. While AI enablers like chipmakers will remain strong, investors are increasingly looking at non-tech companies that integrate AI. Citi's Scott Chronert expects a shift from AI enablers to adopters, leading to more productivity gains across businesses. Goldman Sachs predicts AI will boost S&P 500 earnings growth by 0.4% in 2026 and 1.5% in 2027, with only 30-40% of large firms currently using AI. Sectors like industrials, materials, consumer discretionary, big banks, and pharmaceuticals are expected to benefit most. However, concerns exist that widespread AI adoption could weaken the job market, impacting consumer spending.

Easy AI jobs to start a new career in 2026

A study by 365 Careers identified the ten easiest AI jobs to enter for those seeking a career change in 2026. These roles offer varying salaries and training times, with many having low entry barriers. Top accessible jobs include AI Prompt Engineer, which has seen a 209% increase in UK job postings, and AI Training Data Specialist. Other roles like AI Customer Support Specialist and AI Content Moderator require only weeks or months of training. The list also includes AI Sales, Implementation Consultant, Chatbot Designer, AI Ethics Compliance Officer, Documentation Specialist, and Low-Code AI Developer. These positions offer significant opportunities across Europe, with strong demand in countries like the UK, Germany, and the Netherlands.

Sources

• The Go-To-Market Playbook Isn’t Broken. You’re Just Running the 2021 Version. LIVE with Jason Lemkin
• AI helps me stop thought spirals and return to the real world
• The Most Worrying Bits from Bloomberg’s Worrisome AI Bubble Q&A with Jason Furman
• “Retail Traders Are Just Scratching the Surface of AI,” FMLS:25 Panel on Broker Growth
• How Agility Robotics uses artificial intelligence, from their humanoid ‘Digit’ to everyday workflow
• Indian American-led engineers develop 3D chips to accelerate AI
• U.S. Battles China In AI Race But American Firms Quietly Embrace Cheap, Powerful Chinese Open-Source Models
• AI adoption will play a greater role deciding next year's winners and losers. Here's why
• Career Change in 2026? The “easiest” AI jobs to enter