Venture capital money is flooding into robotics and physical AI startups at an accelerating pace. Investors are backing factory automation and humanoid robot companies in what many experts call a new hardware gold rush. The driving force is a race to collect real-world data and move AI out of the cloud and into the physical world.
Anthropic is preparing for a November IPO but faces serious financial questions. The company reported an $80.7 billion net loss in the first half of 2026, mostly from a $72 billion accounting charge tied to convertible bonds. Despite fast revenue growth, Anthropic faces risks including a narrow customer base and uncertain access to computing power. The firm has promised to spend at least $518 billion over the next decade on cloud services and computing, with new disclosures suggesting total costs could reach $750 billion.
Big Tech AI infrastructure spending is expected to cross $600 billion in 2026 and exceed $1 trillion by 2027, according to UBS. The surge reflects insatiable demand for computing power that continues to outpace supply, putting pressure on company cash flows while supporting broader economic growth.
On the retail front, Walmart and Amazon are taking very different paths on AI. Walmart is rolling out AI agent shopping features to make automated purchasing easier for customers. Amazon, by contrast, is blocking AI bots from its platform. Two retailers, two contrasting strategies for handling AI-driven commerce.
Open-weight AI models are creating new supply-chain risks for enterprises. These models can be downloaded and run outside the original developer's infrastructure, offering lower cost and more control. But they also raise governance questions, since malicious behavior could be baked into the model before it ever reaches a company, and their full characteristics are difficult to inspect.
In other developments, Elon Musk revealed that competing AI companies are eager to access SpaceX computing resources during a recent earnings call, though no specific deals were announced. CaptivateIQ launched two trusted AI agents for sales planning, and the Wall Street Journal introduced an AI tool called Doppler to forecast midterm election issues by analyzing millions of TikTok videos and podcasts.
Key Takeaways
- Silicon Valley is pouring billions into robotics and physical AI startups in a new hardware gold rush
- Anthropic reported an $80.7 billion loss in H1 2026, including a $72 billion accounting charge, ahead of its November IPO
- Anthropic plans to spend at least $518 billion over the next decade on computing, with potential costs reaching $750 billion
- Big Tech AI infrastructure spending is expected to exceed $1 trillion by 2027, crossing $600 billion in 2026
- Walmart is adding AI agent shopping tools while Amazon is blocking AI bots from its platform
- Open-weight AI models pose new supply-chain risks, as malicious behavior could be embedded before companies receive them
- Elon Musk says competing AI companies want access to SpaceX computing power
- OptScientist framework recovers ground-truth optimization laws for over 80% of physics-derived problems tested
- WSJ launches Doppler AI tool to analyze 25 million TikTok videos and 500,000 podcasts for midterm election sentiment
- An AI agent created by UC Berkeley researchers sparked online chaos after being added to a 7,500-person researcher listserv without permission
Silicon Valley pours billions into robotics and AI startups
Venture capitalists are investing billions of dollars in robotics and physical AI companies. Reporter Rya covers this trend for Business Insider, focusing on factory automation and humanoid robots. The surge in funding reflects a race to collect real-world data and bring AI into the physical world. Many investors are joining what experts call a new hardware gold rush.
Elon Musk says AI rivals want SpaceX computing power
Elon Musk revealed during a SpaceX earnings call that competing AI companies are eager to access SpaceX computing resources. The remarks highlighted growing demand for powerful computing hardware in the AI industry. SpaceX has developed significant computing capabilities that could serve AI training needs. Musk did not provide specific details about potential deals or access.
Walmart adds AI shopping tools while Amazon fights bots
Walmart is introducing new AI agent shopping features designed to make automated purchasing easier for customers. The company says these tools will improve the online shopping experience. Meanwhile, Amazon is taking a different approach by blocking AI bots from its platform. The two retailers are choosing contrasting strategies for handling AI-driven shopping.
Anthropic reports huge loss before planned November IPO
Anthropic recorded a net loss of 80.7 billion US dollars in the first half of 2026. This large loss was mostly due to an accounting charge of 72 billion dollars related to bonds that could turn into company shares. The San Francisco company plans to go public in November but investors are now considering a valuation of 1.5 trillion dollars. The firm spent nearly 9 billion dollars on computing capacity and infrastructure in these six months. Anthropic has promised to spend at least 518 billion dollars over the next decade on cloud services and computing.
Anthropic faces risks despite fast revenue growth and IPO plans
Anthropic is growing its revenue quickly but faces many risks as it prepares for a huge initial public offering. CEO Dario Amodei leads the company which has a narrow customer base and uncertain access to computing power. New disclosures show the company might spend 750 billion dollars in coming years to get more computing capacity. The company also has a long list of risk factors that investors need to consider before buying stock. These reports highlight the high costs and challenges in the current AI race.
Open-weight AI models create new supply-chain risks
Companies are using AI models that can be downloaded and run outside the original developer's infrastructure. These open-weight models offer lower cost and more control but create governance questions about how the models were built and how they behave. Unlike regular software, these models contain billions of numerical parameters that can be manipulated by adversaries. This means malicious behavior could be baked into the model before it ever reaches a company. A key concern is that enterprises may import complex reasoning engines whose full characteristics are difficult or impossible to inspect completely.
OptScientist framework discovers optimization laws from data
OptScientist is a new automated framework that discovers compact symbolic optimization laws directly from data. It uses a model that represents data as optimization problems combined with an LLM agentic system for scientific discovery. Tested on 99 physics-derived optimization problems, it recovered ground-truth laws for over 80% of them. It also performed well on human risky choice and central-bank monetary policy data. The framework uses fewer parameters and tokens than LLM-based approaches while achieving competitive predictive performance.
Big Tech AI spending to exceed $1 trillion in 2027
UBS reports that big tech companies will invest over $1 trillion in AI infrastructure in 2027. Spending on AI infrastructure is expected to cross $600 billion in 2026 before reaching that milestone. The growth is driven by increasing demand for computing power that continues to outpace supply. This level of investment also puts pressure on company cash flows while supporting broader US economic growth.
AI model boosts simulated wastewater biorefinery value
A simulated wastewater biorefinery became economically viable using a combination of recurrent neural networks and real options analysis. The AI-enhanced framework produced a net present value of $2.8 million over 10 years compared to negative $1.2 million under conventional analysis. The approach preserves managers' ability to delay or expand investment as conditions change. The study used a Long Short-Term Memory network to predict resource value based on biogas and struvite yields and market prices. Monte Carlo simulation ran 10,000 possible market and operating paths to test the investment model.
Three steps to prepare enterprise data for AI
Preparing enterprise data for AI requires making it searchable, contextual, and trusted. First, companies should use ontologies to map business concepts to data so AI agents can find the right information across multiple platforms. Second, organizations need to provide technical metadata, semantics, ontologies, and knowledge graphs so AI understands the business meaning of data. Third, companies must build trust through governed data, clear access controls, and ownership of definitions and business rules. This work is considered a worthwhile investment tied to business value.
CaptivateIQ launches trusted AI agents for sales planning
CaptivateIQ introduced two new AI tools called Rev Planning Agent and Comp Builder Agent. They help teams set sales quotas, manage incentive pay, and adjust plans faster. The agents work from a company's own data and rules so results are traceable and accurate. Humans must preview every change before it is applied. The tools aim to reduce errors and speed up sales planning decisions.
WSJ launches AI tool Doppler to forecast midterm election issues
The Wall Street Journal created an AI tool called Doppler to study topics shaping the 2026 midterm elections. It has analyzed 25 million TikTok videos and 500,000 podcasts to understand voter sentiment. The tool uses different AI models including LLMs and embedding models to process data. The Journal publishes methodology notes with each story to explain how Doppler was used. Reporters work with engineers to tailor the analysis for each article.
AI agent joins researcher listserv causing online chaos
An AI agent created by University of California Berkeley researchers was added to a 7,500-person listserv of top internet researchers without their knowledge. The listserv is a private forum where scholars share their work. The unexpected appearance sparked debate about privacy and boundaries in online communities. Administrators later removed the agent from the group. The incident raised broader questions about the role of AI in online spaces.
Sources
- Why Silicon Valley VCs are investing billions in robotics and AI
- AI rivals are thirsty for SpaceX's compute, Elon Musk says
- Walmart adding AI agent shopping features while Amazon blocks bots
- Anthropic logged net loss of over US$80 billion in first half of 2026
- Exclusive | Anthropic Draft IPO Filing Shows Red-Hot Growth, Many Risks and Massive Computing Costs
- Open-Weight AI can bring hidden supply-chain risks
- OptScientist: Automated Discovery of Optimization Laws
- AI boom set to push Big Tech investment beyond $1 trillion in 2027, says UBS
- AI and flexible investment model lifts simulated wastewater biorefinery value
- 3 ways to get your data ready for AI
- CaptivateIQ Brings Trusted AI Agents to Incentive and Sales Planning
- The Wall Street Journal’s new AI tool, “Doppler,” forecasts the issues that will shape the midterm elections
- An AI Agent Joined a Listserv of Top Internet Researchers. The Reply-All Chaos Is Still Unfolding
Comments
Please log in to post a comment.