The Biden administration is readying dramatic last-minute steps to preserve a crucial advantage in its AI arms race with China: supply of the world's most advanced chips.

Why it matters: The chips needed to develop cutting-edge AI are the most valuable pieces of hardware on Earth, and the best chips Chinese firms can produce lag about five years behind the top end of the market.

Driving the news: A pending executive order could cap sales of AI chips to countries all over the world, not just China, per the WSJ — with a particular focus on Southeast Asia and the Gulf.

• Biden has already imposed limitations on the advanced chips that companies like Nvidia can export to China, but there are concerns that Chinese firms are able to buy or access them in other countries or from smugglers. There's a thriving black market for Nvidia chips in China.
• The new order would attempt to close that back door. It could also further divide the world along technological lines, with some countries likely getting unfettered access to U.S. tech and others facing limitations.
• Details of the rule, which is pending regulatory review, according to OMB's website, haven't been made public. But U.S. chipmakers and tech firms have been waging an intense behind-the-scenes campaign to prevent more restrictions.

State of the race

The fact that Chinese firms would "opt into a supply chain that involved putting chips in suitcases and smuggling them" is a clear sign of the Western edge in chipmaking, says Chris Miller, a professor at Tufts and author of "Chip War."

• The CEO of one of China's leading AI firms, DeepSeek, said this month that his primary constraint was not the vast sums needed for AI development, but access to high-end chips, Miller notes.

Breaking it down: The chips used to power AI development are mostly designed in the U.S. and fabricated in Taiwan, with chipmaking tools built in the U.S., Japan and the Netherlands.

• Beijing has declared its determination to leapfrog the West in every facet of the semiconductor supply chain. For now, it's locked out.
• The most advanced chips made by SMIC, the largest Chinese chip manufacturer, are on par with the top-end chips Taiwan's TSMC produced five years ago, Miller says.
• The Western advantage in chipmaking tools (such as the extreme ultraviolet lithography machines built by Dutch firm ASML) is so vast that China has little chance of narrowing it over the next five years, says Miller.
• "Everything depends on what type of equipment the West is going to be willing to sell to China. If the restrictions are tight and get tighter, I have high confidence that the West retains its chipmaking lead," Miller contends.

Yes, but: Other experts argue that cutting off access will hamper Chinese firms in the short term, but give them an extra incentive to out-innovate Western competitors in the longer term.

• Beijing is pouring hundreds of billions of dollars into that objective.

The Trump factor

The chip wars have been heating up during the Biden-Trump transition.

• China fired a warning shot earlier this month after Biden's latest export controls were announced by opening an antitrust investigation into Nvidia, causing the $3.3 trillion behemoth's stock to wobble.
• That came a week after China announced it was banning exports to the U.S. of key minerals used in chipmaking.
• The tit-for-tat could continue to accelerate in Trump's second term, given that Biden has been tightening export controls that began during Trump's first term.

The intrigue: While Trump's administration-in-waiting is packed with China hawks, some incoming officials (including Trump himself) have indicated they also want to cut deals with Beijing. One piece of leverage in any such negotiations could be access to chips.

• Meanwhile, Trump's former national security adviser Robert O'Brien argued Tuesday that Biden's looming executive order would "cede the AI market to China" and "drive a wedge between the U.S. and our partners."
• A Trump transition spokesperson did not say whether Trump agrees with O'Brien. The White House referred Axios' questions about the pending executive order to the Commerce Department, which declined to comment.

Beijing's toolbox

For now, China has three main points of leverage:

1. Its massive market:

• U.S. chipmakers like Nvidia and Intel have designed chips for sale to China that fall just within the current regulations, a sign of their intent to continue to fight for market share in China.
• But in addition to the new Nvidia probe, China has announced an investigation into Intel and unveiled a partial ban last year on Micron, the Idaho-based memory chip giant.

2. Its concentration of some of the elementary inputs for chips:

• China cutting off supply of those minerals would have drastic implications for the global economy — but could be hard to execute without hurting China itself, give the global nature of semiconductor supply chains.

3. Its proximity to Taiwan:

• The world's most advanced chips are made almost exclusively on an island around 100 miles off the Chinese mainland, which Beijing has vowed to bring under its control — potentially by military force.
• Taiwan's TSMC claims to make 99% of the world's AI accelerator chips.

What to watch: TSMC is building a fabrication facility in Phoenix and has two more planned in the U.S. as part of a Biden push to onshore semiconductor production.

• For now, though, "all AI progress depends on TSMC production in Taiwan remaining online," Miller says.

President-elect Trump's nominee to lead the National Institutes of Health has drawn scorn for his views on herd immunity and COVID, but Jay Bhattacharya's arrival would put renewed focus on why a research institution with a nearly $48 billion budget doesn't have more breakthroughs.

Why it matters: The controversial Stanford professor could rattle the scientific establishment and turf-conscious lawmakers in Congress, but also satisfy skeptics' calls for a serious look under the hood at how NIH works.

• There's generally less risk-taking today that pushes science in new directions, in part because of economic incentives and the higher likelihood that research confirming earlier work will pay off.
• "Getting science right is arguably the single-most important thing we can do in society," says Caleb Watney, co-founder and co-CEO of the Institute for Progress.

Catch up quick: Trump last month nominated Bhattacharya, a Stanford University health economist, to run the NIH and its nearly 19,000 employees across 27 institutes and centers.

• Bhattacharya was a polarizing figure during the pandemic, criticizing COVID-19 lockdowns and vaccine mandates, and pushing the idea of protecting vulnerable populations like the elderly while letting others resume their lives.
• Ex-NIH director Francis Collins dismissed it as fringe thinking and a diversion from mainstream science. He and others said it was dangerous.
• Collins, who retired at the end of 2021, has more recently expressed some remorse for not considering the full effects of the government's pandemic policies.

What's received less attention is Bhattacharya's work analyzing aspects of NIH funding and calls for revamping the agency.

• He's floated the idea of setting term limits for NIH officials, reevaluating the agency's process for reviewing grant proposals, rethinking how the NIH measures novelty and success so it can take on more risk, and developing other ways to spur and support novel ideas, or what's known as "edge science."
• Bhattacharya "will be a critical asset to bring in needed outside perspectives to re-evaluate the NIH's operations and processes to restore the organization's former gold standard of medical research and better address America's health challenges," Kush Desai, spokesperson for the Trump-Vance transition team, told Axios in an email.

The big picture: Bhattacharya is far from the first to identify how the NIH and other agencies struggle to support risk-taking. (The agency has set up "high-risk, high-reward" programs to support innovative science.)

• And beyond biomedical science, there's an ongoing debate about why the rate of scientific progress appears to be slowing down despite an increase in the number of scientists, amount of funding for their work, and the quantity of papers they publish.
• Some researchers believe the low-hanging fruits of discovery have simply already been plucked, meaning scientists have to work harder and more money needs to be invested to get what remains.
• Others, including Bhattacharya, argue something has gone institutionally wrong with how science is funded — including with taxpayer dollars — and evaluated and are looking for ways to correct the course.

Along with economist Mikko Packalen, his former student who is now an economist at the University of Waterloo in Ontario, Bhattacharya has found NIH is overall supportive of novel scientific ideas.

• But their analysis showed the funding of innovative ideas at the agency skews toward basic science rather than clinical research. They argue the NIH should double down on novel and incremental research in basic science that the private sector doesn't have an incentive to fund.
• They acknowledge, though, that the agency, as a taxpayer-supported institution, is under pressure to deliver results.
• Other researchers have put forth a range of ideas for spurring innovative science, including adopting a model of funding people not projects — an approach taken by the Howard Hughes Medical Institute — or offering prizes as incentives.

Zoom in: Bhattacharya and many others have homed in on the grant review process as an arena for reform.

• Bhattacharya has called for more younger scientists — who have been found to be more innovative than their older counterparts — to be involved in the process.
• Other researchers have suggested changing how grant proposals are scored or using a modified lottery to select grants after those that clearly aren't viable are removed from the pool.
• "Everyone complains about peer review," one former senior NIH official said. But "picking the winners isn't that easy in science."

And any changes could be slow in coming since the NIH director doesn't have the undisputed final word.

• Committees at the NIH and across agencies are bound by the Federal Advisory Committee Act, or FACA, of 1972.
• Changes instituted under former President George W. Bush's NIH director Elias Zerhouni in 2006 took an act of Congress.

Others argue the root of the risk-taking problem runs far deeper and beyond the bounds of the NIH.

• Harold Varmus, who led the NIH under former President Clinton, says the real problem is hypercompetition. "Too many good people are looking for support and the result of high-level competition is a tendency to look for the most secure ideas to pursue."
• "When the success rate of a grant being funded is 10% or 20%, scientists look for a safe route."

Between the lines: Another big question is how much high-risk science the NIH — or any agency's — portfolio should have, and what level of risk is acceptable for public funding.

What to watch: Bhattacharya's supporters — and those who expressed cautious optimism about his nomination — say he could run some experiments to test these ideas in the high-risk, high-reward programs where the director does have more direct discretion.

• "There is so much to do. None of this glamorous," says Pierre Azoulay, an economist at MIT who was a co-investigator on a grant with Bhattacharya before the pandemic.
• He says its also important not to lose sight of the fact that the NIH "still funds a lot of great research" and "you can imagine some reforms doing more harm than good."
• But he says, "it leaves me hopeful that there is a nominee ... who has thought about the scientific enterprise using the scientific method — that is a complete first."

Editor's note: The story has been updated to include Collins' more recent reflections on the COVID pandemic response.

The world's two richest men and biggest space entrepreneurs both believe humanity's survival depends on life beyond Earth — for very different reasons.

• Blue Origin founder Jeff Bezos explained this week that his vision is to move all polluting industries into space to preserve Earth.
• His rival Elon Musk envisions inhabiting space as a way for humanity to live on beyond Earth, if a cataclysm strikes our planet.

Why it matters: Both are pouring tens of billions of dollars into space travel with those endgames in mind.

• "These are visions of potential futures, and there is a long gap between achievement and vision," John Logsdon, founder of the Space Policy Institute at George Washington University, tells Axios.

Bezos' vision is for Blue Origin to lay the groundwork such that "the next generation, or the generation after that, will be able to move polluting industry off Earth, and then this planet will be maintained as it should be," he said Wednesday at the NYT's DealBook conference.

• In Bezos' view, human civilization needs to continually use more and more energy in order to advance, but Earth's resources are finite and must also support many forms of life beyond humans. That means tapping natural resources beyond Earth.
• "There is no plan B. We have to save Earth. We've sent robotic probes to all of the planet solar system. This is the good one, and we must save it," Bezos said.
• Bezos has also previously spoken about his dream of building enormous space stations (known as O'Neill cylinders) in relative proximity to Earth's orbit, to allow people to travel back and forth.

Musk's SpaceX, meanwhile, is actively drawing up plans for what life could look like on Mars, including specialized spacesuits and domed habitats, per the NYT.

• Musk has repeatedly argued that a mass extinction event on Earth, such as an asteroid strike, is inevitable. "Either become a spacefaring civilization or die," he posted on X in September. "Those [are] the two choices."
• Ultimately, he envisions transforming Mars to turn the Red Planet green (one tool he has floated: nuclear detonations).
• He has admitted in the past that developing a martian civilization could take centuries, but reportedly claimed earlier this year that one million people will move to Mars within two decades.

Reality check: Few believe that timeline is anywhere close to realistic. Even in the longer term, "I don't know why a million people would want to go to Mars," Logsdon says. "What would the economics of that be?"

• Other experts have noted that when pondering the survival of the species, preserving human life on Earth is quite a bit simpler than developing it on Mars.
• The obstacles to Bezos' vision of a network of colossal space stations — such as maintaining a supply of food and water in an entirely artificial environment — are hardly less daunting.

Between the lines: Both billionaires could make an astronomical amount of money from their space ventures even if their dreams of building space civilizations don't come to fruition.

• SpaceX is considering selling shares at a whopping $350 billion valuation, Bloomberg reported this week. That's because its Starlink satellite system and Starship mega-rocket — developed to fund and power Musk's Mars ventures — are already so valuable to governments and militaries here on Earth.
• Blue Origin is growing more slowly, but Bezos argued Wednesday that it will ultimately be more profitable than Amazon.

LONDON — Top researchers this week said scientific discoveries using AI, like new drugs or better disaster forecasting, offer a way to win people's trust in the technology, but they also cautioned against moving too fast.

Why it matters: Public trust in AI is eroding, putting the technology's wide adoption and potential benefits at risk.

Driving the news: At a forum in London hosted by Google DeepMind and the Royal Society, a roster of renowned scientists described how AI tools are transforming and turbocharging science.

• Efforts range from the search for beneficial new materials to the quest to build a quantum computer to the potential for self-driving labs.

Between the lines: In industry, the buzz around AI has largely centered on the technology's capacity to streamline business — along with the possibility that it might advance toward artificial "superintelligence."

• Experts at the London event highlighted AI as a scientific tool and argued that the scientific method will best serve researchers seeking to leverage advanced AI models and fathom their complexity.

But, the painstaking, thorough work of science can be at odds with the "move fast break things" ethos of the tech industry that is driving AI's development.

• Scientists in the U.S. also face a tide of skepticism about their work.

What they're saying: "I think the scientific method is, arguably, maybe the greatest idea humans have ever had," DeepMind CEO Demis Hassabis told the London gathering.

• "More than ever we need to anchor around the method in today's world, especially with something as powerful and potentially transformative as AI," he said, adding that he thinks neuroscience techniques should be used to analyze AI "brains."
• "I feel we should treat this more as a scientific endeavor, if possible, although it obviously has all the implications that breakthrough technologies normally have in terms of the speed of adoption and the speed of change."

Zoom in: A slew of recent papers show how scientists are trying to put AI to work on some of nature's most complex problems.

• Scientists from the Arc Institute built an AI model trained on the DNA sequences of microbes rather than words and sentences of text.
• This "genomic foundation model" can predict how a DNA change affects an organism and generate realistic genomes from scratch, which could one day help scientists to engineer biology with more ease and precision.
• Researchers have also developed an AI model of the chemical modifications that turn genes on and off.

An ambitious AI-driven effort is underway to map the human body's 37.2 trillion cells.

• It has yielded discoveries — including insights into the development of the human skeleton and the immune system — that were published last week in more than 40 papers from the Human Cell Atlas consortium.

"It's just dizzying. I've never seen anything like it in my life," Eric Topol, founder and director of Scripps Research Translational Institute, said at the event.

Yes, but: "We're moving so fast, we've got to be careful," said Alison Noble, a professor of biomedical engineering at Oxford University.

• "It's great to hear about all the excitement" around AI, she said, but researchers in the field need to re-commit to the basics of the scientific method, like being able to reproduce results from experiments.
• Scientists have expressed concern that some AI tools are being used without understanding the nuances of their abilities and their limits — and creating a reproducibility crisis that could undermine trust in the both the science and the tools.

There also needs to be a shift in how AI-enabled discoveries are described, Denis Newman-Griffis of the Centre for Machine Intelligence at Sheffield University told Axios.

• Statements like "AI discovered new protein structures" ignore that people designed the algorithms, chose the data to train models, interpreted the AI's output, and "built the entire research system those tools are operating in," they said.
• "[W]e cede all the agency that we have" and paint a picture of AI as "nebulous, difficult to control, impossible to understand, and so directly opposite to the things that would make its use trustworthy."

The big picture: Google's top executives in attendance — Hassabis and James Manyika, senior vice president of research, technology and society — said they're trying to increase trust in AI by using it to solve practical problems, including forecasting floods and predicting wildfire boundaries.

• "What could be a better use of AI than curing diseases? To me, that seems like the number one most important thing anyone could apply AI to," Hassabis said.

What's next: Next month, Hassabis and his colleague John Jumper will collect their Nobel Prize for developing AlphaFold, an AI system that can predict the structure of proteins and is used for drug discovery and other problems.

• The challenge "is a lot of those things, as useful as they are, people may not immediately think of them as AI," Manyika said.