White House: US must dominate robotics, too

After several months of high-profile dealmaking and promotion of a plan for US dominance of AI, the Trump White House is expanding its sights to include dominance of robotics as well. Commerce Secretary Howard Lutnick is reportedly meeting with the CEOs of several US robotics companies, and says he is “all in” on accelerating the industry’s development in this country.

This is another example of Trump’s embrace of Chinese-style industrial policy in order to compete with China. There are 2 immediate hurdles to be overcome for applying this model to the robotics sector. The first is that accelerating robotics is likely to clash with Trump’s stated goal of bringing manufacturing jobs back to the US. Robots are usually seen as substitutes for human labor. So, a strong robotics industry may help reshore manufacturing, but may not increase, or may even decrease jobs in US factories.

The more significant obstacle to US dominance of robotics is that China is hands down both the dominant manufacturing power in the world, as well as far and away the largest user of industrial robots. China accounts for 80% of annual installations of industrial robots, and has 43% of the world’s total existing installations, while the US currently accounts for only 7% of new installs annually, and holds only 12% of total installations.

The upshot is that while the US holds a small, and steadily diminishing, lead over China in AI, China currently has a commanding lead over the US in robotics. US dominance in robotics is, um, a decidedly “ambitious” goal.

After AI push, Trump looks to robotics

China dominates both manufacturing and the use of industrial robots

Google tells advertisers that its AI will have ads in 2026

Ad-free AI may soon be a thing of the past. Not even a week after OpenAI caused a flurry of backlash by trying to sneak ads into some answers given to users with paid subscriptions, Google has reportedly been telling advertising clients to expect a rollout of ads attached to its AI answers.

Google’s entire empire is primarily fueled by the advertising it sells in its dominant Search product. The trouble is, AI answers to date have not included ads, and few people click on the links provided in the AI-generated summary.

Since Google and other AI model makers are spending hundreds of billions on AI infrastructure, and only reaping single billions in revenue, at some point in the near future revenue has to increase. Ads are a proven route to vast revenue, so it is likely that sooner rather than later, ads will become standard in at least all free subscriptions.

Google tells advertisers that its AI will have ads in 2026

Trump: Nvidia can sell chips to China if US gets 25% of revenue

Former President Joe Biden banned export of America’s most advanced AI chips to China, citing national security grounds, and until now President Trump has maintained the ban. Top AI chipmaker Nvidia suffered a catastrophic loss of market share in China, going from a high of 95% of the market in China to 0%. Nvidia CEO Jensen Huang has ever since been energetically lobbying the US government for a relax of this ban, and this week Trump posted an announcement on his Truth Social online media site that he had struck a deal with Huang and other US-based chipmakers.

Apparently, US chipmakers will be able to sell not their best and most advanced chips to China, but their second-best chips, so long as they pay 25% of the revenue collected to the US government.

This unusual pay-to-play policy has raised eyebrows, but is perfectly on-brand for Trump, who considers himself Deal-maker in Chief.

One slight complication to this turnabout is that the Chinese government, faced with the export ban, decided to take its fate into its own hands and accelerate chip development in China. In addition, inclusion of foreign chips in products supported by the government, such as the military and any government agency, was expressly forbidden.

In the end, the export ban that was created to hinder China’s progress in AI may have only accelerated it.

Trump’s pay-to-play export policy on AI chips

Robo-taxis are spreading across the US

Once a novelty confined to a few test sites, driverless taxicabs are now rolling out to cities across the country (see map below.) Robo-cabs are already operating in San Francisco, Los Angeles, Phoenix, Austin Texas, and Atlanta. Ten more cities are targeted for rollout in 2026, with eight more beyond that in the planning stages. The leading driverless cab company is Google-backed Waymo, but Amazon’s Zoox, Tesla, and Uber (allied with multiple self-driving technology companies) are moving aggressively into the market. Once riders become convinced of the safety of the concept (and data so far indicates that driverless taxis are far less likely to be involved in an accident than human drivers) many come to prefer it, as it avoids the forced intimacy with sometimes sketchy drivers that is inherent in the current taxi/ride-hailing industry.

Robo-taxis are spreading across the US

OpenAI’s State of Enterprise AI 2025 report

OpenAI has released its 2025 report on the state of AI in large enterprises. The overall message is that enterprise use of AI is growing briskly and is beginning to deliver measurable results in productivity and revenue enhancement. The largest growth in usage has been for tasks that require computer coding. Perhaps the most interesting result of their analysis is that employees who become skilled in using AI tools are starting to take on and accomplish tasks that they never had the time or the skills to do before. A major example of this trend is employees using AI to help them code a mini-app or an Excel spreadsheet that solves an annoying workflow issue. Given the potential usefulness of being able to get your computer to perform almost any function you want just by talking to it, AI may make computer programmers of us all.

OpenAI’s State of Enterprise AI 2025 report

UK robotics company crams 19 months of training into 48 hours

Humanoid, a robotics startup based in the UK, has just released their first humanoid robot, HMND 01 Alpha Bipedal. The robot is too new to have much of a performance history, but it has already set some records during production. First, the Humanoid team were able to move from initial design to working prototype in only 5 months, as compared to a more usual 18-24 months, by repurposing some design aspects of their existing Alpha series of wheeled robots. Second, they were able to train up this first version to walk stably in only 48 hours, by speed running the robot brain through a training database of 52.5 million seconds (equivalent to 19 months, 24/7) of simulated robot locomotion data in only 2 days. The database is supplied by Nvidia as part of their effort to jumpstart robotics development, in order to sell more chips for robot brains.

UK robotics company crams 19 months of training into 48 hours

MIT engineers design an aerial microbot as agile as a bumblebee

MIT has developed a tiny flying robot with insect-like speed and agility, for use in search, rescue, and reconnaissance missions. The tiny robot’s motions are controlled by an AI system that adapts and learns how to produce complex motions, even in rapidly changing conditions. In one test, a microbot was able to complete ten aerial flips in 11 seconds despite gusting winds. Currently the AI flight controller is housed on an external computer. The team is now working on miniaturizing the controller in order to place it on the microbot itself, allowing full autonomy of operation.

MIT engineers design an aerial microbot as agile as a bumblebee

UCLA research leads to first robot-assisted cataract surgery

A medical engineering team at UCLA has developed a new robotic surgery system that has enabled the world’s first robot-assisted cataract surgery. Surgery on the eye requires high levels of precision, beyond the abilities of existing robotic systems such as the da Vinci system for soft tissues (such as gall bladder or prostate surgery), or the Stryker system for orthopedics. The UCLA system has shown tooltip accuracy of 0.053 millimeters in laboratory testing, approximately the diameter of a microscopic grain of pollen.

UCLA research leads to first robot-assisted cataract surgery

8VC’s vision for AI in Health Care

Venture capital firm 8VC has released a white paper on its vision for AI in health care. The paper starts with an indictment of current US health care as a system that is broken in multiple ways. The costs are too high, insurance premiums are spiraling, seeking and receiving care is needlessly complex and inefficient, wait times for doctors appointments now averages 31 days and is rising, physicians have no time to carefully explain issues of their patients’ health to them, and physicians themselves are burning out due to “the drudgery of button clicks.”

Their solution? A massive transformation of the healthcare system by aggressive infusion of AI. AI can be the ever-available chatbot guiding the patient through the health care maze, explaining their symptoms, the results of their tests, and their available; treatment options. AI can act as a personalized health coach for each patient. It can automate administrative tasks, and gradually become ever more helpful in diagnosis and treating patient diseases. 8VC can even envision a day when AI systems can diagnose and treat patients autonomously. Sort of like a Waymo robotaxi, but for health care.

They are curiously silent on the possibility of AI taking over the roles of venture capitalists.

8VC’s vision for AI in health care