Deploying AI agents to accelerate scientific
discovery is an exciting prospect. The U.S. and China, the world’s leading countries
for scientific research and AI, are in competition. Both have recently launched
agentic AI networked systems to accelerate data analysis and discovery. These systems,
the Genesis Mission of the U.S. and China’s AI+ Initiative, will utilize and
network powerful supercomputers.
The
U.S. Genesis Mission
The Genesis Mission was announced by President
Trump on November 24, 2025. It is part of the Department of Energy.
According to Wikipedia:
“The
mission seeks to create a centralized AI platform designed to accelerate
breakthroughs in various fields, including space exploration, healthcare, and
national security. The initiative's overarching goal is to harness AI's
potential to solve complex global challenges and enhance the nation's
technological competitiveness. Through public and private sector collaboration,
Genesis Mission 2025 will focus on developing advanced AI models that can
support major scientific endeavors, including space exploration and
environmental sustainability.”
According to Sandia Laboratory News, one
goal of the Genesis Mission is to “double the productivity and impact of
American science and engineering within a decade.” The competitive aspect
of the project is revealed by the use of terms like “technological dominance.”
“The
Genesis Mission will mobilize the DOE’s 17 national laboratories, industry and
academia to build an integrated discovery platform.”
“The
platform will connect the world’s best supercomputers, AI systems and
next-generation quantum systems with the most advanced scientific instruments
in the nation. Once complete, the platform will be the world’s most complex and
powerful scientific instrument ever built. It will draw on the expertise of
roughly 40,000 DOE scientists, engineers, and technical staff, alongside
private sector innovators, to ensure that the United States leads and builds
the technologies that will define the future.”
An article by Kaif Shaikh in Interesting
Engineering explains the competition and the capabilities being pursued:
“…scientific
leadership in the twenty-first century will be determined by who can best
combine data, computation, and algorithms into one continuous cycle of
learning, experimentation, and innovation.”
“AI can
process far more data than humans, test millions of possibilities, and uncover
patterns that researchers would never see in time. Deep Learning systems can
absorb the entire known record of a scientific field, learn the underlying
rules, and then use that intuition to design new molecules, materials, or
reactor configurations. By applying these techniques directly to the nation’s
scientific infrastructure, the Genesis Mission aims to give every major
research area access to the same capabilities.”
The Genesis Mission aims to create:
“…the
American Science and Security Platform, a secure, national-scale research
system that integrates DOE laboratory supercomputers, cloud-based AI computing
environments, scientific models, and extensive federal datasets.”
According to the executive order, the goal of the American Science and Security Platform is to provide:
(i) high-performance computing resources,
including DOE national laboratory supercomputers and secure cloud-based AI
computing environments, capable of supporting large-scale model training,
simulation, and inference;
(ii) AI modeling and analysis frameworks,
including AI agents to explore design spaces, evaluate experimental outcomes,
and automate workflows;
(iii) computational tools, including AI-enabled
predictive models, simulation models, and design optimization tools;
(iv) domain-specific foundation models across the
range of scientific domains covered;
(v) secure access to appropriate datasets,
including proprietary, federally curated, and open scientific datasets, in
addition to synthetic data generated through DOE computing resources,
consistent with applicable law; applicable classification, privacy, and
intellectual property protections; and Federal data-access and data-management
standards; and
(vi) experimental and production tools to enable
autonomous and AI-augmented experimentation and manufacturing in high-impact
domains.
Shaikh points out in his article that AI
has already accelerated many scientific discoveries so the likelihood of a more
powerful networked system should be able to exceed previous discoveries. Some
of the fields where it could yield results include manufacturing,
biotechnology, critical materials, nuclear fission and fusion, quantum
information science, and semiconductors. The networking aspect connects the
national labs, academia, and industry.
The U.S. also has a large private sector
AI initiative, in collaboration with the U.S. government, known as the Stargate
Initiative. Initial funding commitments are $100 billion, with plans to spend
$500 billion within four years. Much of the initiative involves AI
infrastructure buildout in the form of advanced AI data centers. It was announced
in January 2025. It was also reported that OpenAI was pitching the construction of some
data centers as large as 5 GW in power consumption. That is pretty immense for a
single facility.
China’s
AI+ Initiative (and the BIE-1 Portable Supercomputer)
In late October 2025, China announced that
it had developed a refrigerator-sized supercomputer, the BIE-1, that could rival
much larger models with 90% less energy use. The feat was said to be achieved
through the use of an intuitive neural network and a brain-like AI algorithm. According
to Interesting Engineering:
“The
BIE-1 integrates 1,152 CPU cores with 4.8 terabytes of DDR5 memory and 204 TB
of storage space. It closely mimics the human brain’s computational mechanisms,
enabling efficient learning and interpretable reasoning. It also displayed the
ability to learn and extract patterns from small data.”
It achieves training speeds of 100,000
tokens per second and inference speeds of 500,000 tokens per second, which
rival much larger supercomputers. One goal is to provide access to supercomputer
power for non-professionals.
In December 2025, China announced that it
had networked its system of supercomputers across 1,243 miles, with 98% of the energy
efficiency of it being a single data center. The link-up is part of the Future
Network Test Facility (FNTF), China’s first major national infrastructure
project in the information and communication sector, launched a decade ago, but
now ready to operate. The network utilizes optical transmission lines. I assume
this is the same as what we call fiber-optics or optical fiber technology.
“Its
capabilities were demonstrated at last week’s launch ceremony, when a
72-terabyte dataset from FAST, the world’s largest single-dish radio telescope,
was transmitted across 621 miles in just 1.6 hours. Over the regular internet,
the transfer would have taken about 699 days.”
The network consists of more than thirty
supercomputers and is known as the National Supercomputing Network, or SCNet.
According to Interesting Engineering’s Sujita
Sinha:
“According
to China Science Daily, the new AI agent can accept simple natural-language
instructions and carry out entire research workflows with minimal human
involvement. Once given a task, it can break the problem into steps, allocate
computing resources, run simulations, analyze large datasets, and generate
comprehensive scientific reports.”
“The
system is designed to function as a self-directed research assistant rather
than a passive tool. Officials say this approach dramatically reduces the time
required for complex scientific work. Tasks that previously took researchers a
full day can now be completed in about an hour.”
“Nearly
100 scientific workflows are currently supported. These span key areas such as
materials science, biotechnology, and industrial artificial intelligence. The
focus is on accelerating discovery by automating routine but computationally
intensive research processes.”
The
Future? It May Arrive Sooner with Armies of AI Agents as Digital Assistants and
Linked Supercomputers on the Job
It looks like the U.S. and China are
expanding AI, including agentic AI, at roughly a similar rate. It is, however,
difficult to compare the two countries’ efforts without real data for
comparison. It bodes well for interesting times in scientific discovery ahead. Along
with discovery, I would assume AI will continue to provide optimization of many
processes and systems.
References:
China
rolls out super AI science network to challenge Trump’s Genesis Mission. Sujita
Sinha. Interesting Engineering. January 1, 2025. China
rolls out super AI science network to challenge Trump’s Genesis Mission
Genesis
Mission: A National Mission to Accelerate Science Through Artificial
Intelligence. U.S. Department of Energy. Genesis
Mission
Genesis
Mission. Wikipedia. Genesis
Mission - Wikipedia
DOE
launches Genesis Mission to transform American science and innovation through
AI computing revolution. Darrick Hurst. Sandia Lab News. December 4, 2025. DOE
launches Genesis Mission to transform American science and innovation through
AI computing revolution – LabNews
The
Genesis Mission: How the US plans to rebuild scientific discovery with AI: Genesis
boosts energy, biotech, materials, and national security. Here’s how it works. Kaif
Shaikh. Interesting Engineering. November 28, 2025. What
is Genesis Mission, and how it speeds up US scientific research
China
activates massive distributed AI system spanning 1,243 miles nationwide. China
switches on a nationwide optical backbone that fuses scattered data centers
into an ultra-fast, unified AI supercomputer. Neetika Walter. December 11,
2025. China
activates 1,243-mile distributed AI supercomputer network
China
unveils ‘fridge-sized’ AI server that slashes power use by 90%: The
breakthrough design aims to make advanced AI computing more sustainable and
accessible for businesses and research labs. Atharva Gosavi. Interesting
Engineering. October 27, 2025. China’s
compact AI server claims 90% lower power consumption
LAUNCHING
THE GENESIS MISSION. Executive Orders. The White House. November 24, 2025. Launching
the Genesis Mission – The White House
Trump plans $500 billion AI power to defeat diseases, enemies at ‘unprecedented’ rate: OpenAI CEO Sam Altman called the venture ‘the most important project of this era.’ Sujita Sinha. Interesting Engineering. January 22, 2025. Stargate: US plans $500 billion AI project to defeat deadly diseases
No comments:
Post a Comment