Alphabet-owned robotics company, Intrinsic, is joining Google to unlock the promise of physical AI for a much broader set of manufacturing businesses and developers. It will remain a distinct entity within Google but work closely with Google DeepMind and tap into Google's Gemini AI models and cloud services. Intrinsic announced a joint venture with Foxconn in October that involves the two companies working together on general-purpose intelligent robots to transform how electronics are manufactured, with the goal of full factory automation. The company is now working towards those goals with closer collaboration with Google's AI prowess.
Wayve has raised $1.2 billion in a deal that values the London-based autonomous vehicle software startup at $8.6 billion. The funding, which could grow to $1.5 billion if it hits certain performance targets, is one of the largest amounts raised by a European startup. Wayve expects to have its driverless taxis on the road for customers this year through a commercial trial with Uber. Consumer vehicles incorporating its technology will be available to buy by 2027.
Software contracts, tests, and API surface area are becoming more valuable as AI becomes better at cloning people's work. This clashes with the incentives of commercialized open source work. Previously, the overhead of writing, understanding, and maintaining the code itself was important. However, clearly defining and communicating open source software's APIs makes it easier for AI to clone work.
Several things are happening in software: a lot more startups with super-fast revenue growth are popping up, margins are thinner, engineering inequality is growing, and digital natives had a bunch of layoffs. This is all a direct corollary of the platformization of software, and it will continue. What's happening is expected - it just took some time to happen. The software industry is now much bigger than it was a decade ago, and AI is creating a ton of FOMO.
AI increases productivity on well-scoped tasks where errors are detectable. This breaks down when tasks get harder. When AI agent output is scored algorithmically, it looks moderately capable, but when scored holistically, performance drops substantially. Estimates about AI's effect on productivity drop when task reliability is factored in.
vinext is a drop-in replacement for Next.js built on Vite for about $1,110 in tokens that builds production apps up to four times faster and produces client bundles up to 57% smaller.
The US economy built a giant rent-extraction layer on top of human limitations over the past fifty years, with trillions of dollars of enterprise value depending on increased friction.
As the first engineer on TLDR's new Applied AI team, you'll build AI agents and composable Claude Skills to let non-technical teammates create their own AI workflows. Learn more [19].
Most talented developers do not appreciate the impact of the intuitive knowledge they bring to their coding agents, and most of the work people are sharing are incredible personal tools, but they are not capital-P products.
You want AI to be more than a never-ending work in progress, and that means enabling no-code, low-code, and pro-code development without IT gatekeepers standing in the way. Yet that doesn't mean governance goes out the window.
Anthropic has until Friday to comply with the Pentagon's demands to use its AI models for all lawful use cases, including mass domestic surveillance or autonomous weapons, which Anthropic doesn't allow its models to be used for. Failing to comply will mean the cancellation of Anthropic's contract. Defense Secretary Pete Hegseth has said he could also label the company a supply-chain risk or invoke the Defense Production Act to essentially force Anthropic to work more collaboratively with the Pentagon. Designating a company as a supply-chain risk would mean that businesses working with the Pentagon would have to certify that they don't use Anthropic's models in any of their work with the military.
Meta has agreed to buy 6 gigawatts' worth of AI computing power from AMD in a deal valued at more than $100 billion. AMD has agreed to give Meta warrants to buy up to 160 million AMD shares, roughly 10% of the company, for $0.01 apiece, as long as certain milestones are met. Meta said last week that it would buy several million of Nvidia's GPUs as well, a deal expected to cost tens of billions of dollars. The company plans to deploy tens of gigawatts of data center computing power this decade, and hundreds of gigawatts or more over time.
Taara, a Google moonshot spin-off, plans to create urban data highways made out of light. Taara Photonics is a wireless communication platform based on optical phased arrays. The Taara Beam transmits data wirelessly with Class 1M eye-safe lasers in the unlicensed infrared band 193 THz. It can support speeds of up to 25 Gbps over a claimed range of up to 10 kilometers. The Taara Lightbridge Pro can do double that distance.
Over the past month, six American companies and a Chinese firm have expressed interest in building data centers in space. Experts warn that the projects could open governance loopholes, especially for countries with little say in how they are run. Orbital data centers could place critical infrastructure beyond regulatory reach for many developing countries and deepen digital dependence. New frameworks need to be built to prevent orbital compute from becoming an extension of existing terrestrial monopolies.
The heaviest users of AI in fintech aren't fintechs, they're fraudsters. If you watch Plaid's 2026 Fintech Predictions [9], you'll see why that won't change for a while. Plaid CEO Zach Perret, CTO Will Robinson, and Credit Product Lead Michelle Young share their
This post takes a look at the many interesting coding projects that developers, creatives, and entrepreneurs in China have been creating with AI tools over the past year. Tinkering culture has no borders, and companies are cashing in. Vibecoding is changing the landscape of technology in China. Highlights include a lighting app that became the most downloaded paid iPhone app in China, an app that suggests meet-up locations, and a tool that generates elaborate greeting messages based on occasion, style, and relationship with the intended recipient.
November was an inflection point with AI coding. Everyone now wants their own custom app to solve their business problem, and it's easier than ever to create one. Projects that should take weeks are being one-shotted by Claude Code. Developers should take advantage of this moment - in a few years, motivated people will be able to just do it themselves.
Stripe is reportedly considering buying all or some segments of PayPal's business. PayPal has struggled with growth in an increasingly competitive financial payments industry. Meanwhile, Stripe hit a $159 billion valuation on Tuesday, up from $91.5 billion a year ago. Stripe's co-founder and president, John Collison, said that the company isn't yet aiming for an IPO as that would sidetrack its current product and business growth.
Some tech firms have started tracking their workers' use of AI tools with an eye on productivity gains, and in certain cases, factoring it into performance reviews. There are jobs that aren't even considering candidates unless they demonstrate AI fluency. AI adoption hasn't been easy, as tech workers have many of the same feelings about AI as the broader population, including skepticism about how much time it is actually saving them. There's also the added anxiety about adopting a technology that will cause them to eventually lose their jobs.
The AI community tends to say 'this is the worst it will ever be' in response to criticism, but in many aspects, this is also the best it will ever be.
As the first engineer on TLDR's new Applied AI team, you'll build AI agents and composable Claude Skills to let non-technical teammates create their own AI workflows. Learn more [19].
Anthropic claims that DeepSeek, Moonshot, and MiniMax used approximately 24,000 fraudulent accounts to train their own chatbots. The process of distillation, where data from one AI system is used to train another, is common in the AI industry. However, Anthropic's terms forbid the practice and also do not allow its technologies to be used in China. Anthropic has called on government officials and other AI companies to help prevent Chinese companies from distilling American models.
SpaceX, OpenAI, and Anthropic are set to IPO at an unprecedented scale. Typical IPOs offer 15% to 25% of their shares to public markets, which creates enough liquidity for price discovery while allowing founders and early investors to maintain control. At standard float percentages, these three companies need to raise $432 billion to $576 billion from public markets in a single quarter. The entire US IPO market raised $469 billion between 2016 and 2025. It is likely these companies will debut with smaller floats.
The biggest projects in geothermal started construction around 18 months ago. The industry executed well enough to get a shot at true commercial-scale deployments. The next few years will determine if the wave of projects can lead to the success metric of '10% of electricity generation'. While the first version of the technology seems possible, subsequent versions and improvements that lead to scalability will be much more challenging. The leading companies are still going strong, which is promising.
Attention has started turning to using AI to probe the possibility of digitizing smell. Researchers hope that the ability to program smell will illuminate the many mysteries of olfaction. Digital scent could have many practical and quite profitable applications. This article looks at the state of the technology and introduces the various possible avenues for research.
The line between technical and non-technical work is being redrawn. The winners will be the ones who think like builders. Non-technical people should try to take on more 'engineering' tasks as AI makes tools more powerful and accessible. Engineers should use the same tools to turn themselves into full-stack shipping machines. Companies are investing billions to make engineers more powerful, so take advantage of that investment.
Coding agents have dramatically dropped the cost of producing code. The ability to run parallel agents has compounded this effect. However, delivering good code still has a cost. Developers still need to ensure that the produced code is good code. The challenge is to develop new personal and organizational habits to respond to the new reality of agentic engineering.
A recently published paper called 'The 2028 Global Intelligence Crisis' explores a hypothetical world in which there is an aggressive AI build-out and adoption of AI agents. In this world, unemployment is high despite elevated nominal growth and productivity, and the stock market tumbles due to credit stress from companies laid low by AI metastasizing. It illustrates how AI becoming good could leave many industries vulnerable. This post provides a summary of the paper along with a link to the original.
Many jobs in tech aren't that rewarding anymore. AI has made it even worse. However, leaving your current job without a plan right now isn't a good idea because every CEO wants to get leaner. Companies want to increase AI use rather than hire more people, so the scales are weighted against employees. Use the time in your current job to learn and grow more until you have a better place to go to.
12 months of real engineering data reveal what actually changes with AI adoption — speed, stability, and measurable outcomes. Download the practical, data-backed AI measurement guide for engineering