Summary of Key Points
This article uses the scientific concept of "simulated annealing" to compare the two innovation models of China and the United States: The U.S. operates in a "high-temperature state" – allowing chaos and tolerating failure in pursuit of breakthroughs from scratch (0 to 1), while China operates in a "low-temperature state" – focusing on efficiency and stable delivery, excelling at taking something from 1 to 100. However, the boundaries between the two models are becoming blurred. China is now engaging in innovation from 0 to 1 (such as developing open-source model architectures and original innovative drugs), and the U.S. is working on taking something from 1 to 100 to perfection (such as reusing rockets to reduce costs). At the same time, the U.S.'s "high-temperature model" is facing challenges in areas like immigration policies and capital expenditures, and it is uncertain whether it will be able to continue to leap towards new levels of innovation in the future.
Detailed Analysis
1. Innovation Models through "Simulated Annealing": High Temperature Burns Chaos, but Also Burns Through Breakthroughs
"Simulated annealing" is originally a metallurgical process where metal is heated and then cooled, causing atoms to rearrange in a more orderly manner. Computers use this concept to find the "optimal solution": if they always move in the steepest direction at each step, they will get stuck in local optima; by temporarily allowing for "chaos" (high-temperature fluctuations), they can find the global optimum.
- The U.S.'s "High Temperature": The U.S. tolerates chaos (such as temporary failures and disruptions in markets or projects like the five explosions of SpaceX's Starship) and has produced breakthroughs like GPT-5.6 from OpenAI and frequent space launches from SpaceX. The cost is high in terms of money and order, but it allows for the possibility of discovering major opportunities.
- China's "Low Temperature": China has succeeded in making its open-source models the most downloaded in the world (over a billion times) and its innovative drugs account for 46% of the global market. The benefits are high efficiency and low costs, but it sacrifices some of the "crazy" experimentation.
2. The AI Race: The U.S. Sets the Direction, China Builds the Free Foundation
The AI landscape is no longer dominated by the U.S.; instead, there is a division of labor:
- The U.S. Leads the Frontiers: Companies like OpenAI and Google define the next steps in AI development (e.g., naming GPT-5.6 versions after planets in the solar system), holding the "steering wheel of innovation."
- China Builds the Foundation: China accounts for the majority of the top ten open-source models globally, and 80% of AI startups in Silicon Valley use Chinese models. This turns previous innovations into free public goods, but the question of who benefits from these free foundations remains unresolved in history.
3. Innovative Drugs: China Creates the "Molecules," the U.S. Controls the "Pricing Power"
These statistics are counterintuitive:
- China's Progress: By 2025, 46% of the new molecules entering human trials globally will come from China (up from 17% ten years ago); China also holds 90% of the global authorized transactions in the ADC (Antibody-Drug Conjugate) sector, with total external authorizations in the first half of 2026 reaching $99.7 billion (double the annual total for 2024).
- The U.S.'s Advantage: The U.S. buys 42.9 billion out of the $99.7 billion in authorized transactions, demonstrating its dominance. Its advantage lies in its ability to set prices through the dollar, FDA reviews, and global distribution networks. However, if other countries develop their own review and distribution systems, this advantage may disappear.
4. Spaceflight and Nuclear Fusion: Failure is Not a Cost, but a Means of "Sampling"
The U.S.'s tolerance for failure is a core aspect of its "high-temperature model":
- SpaceX's Starship: Despite 12 attempts and 5 failures, the project continues because failure serves as a means of "sampling" to determine the limits of what is possible. If evaluated by traditional KPIs, the project would have been abandoned long ago.
- Nuclear Fusion: Sam Altman's investment in Helion achieved fusion at 150 million degrees Celsius, and TAE Technologies merged with a social media company and went public (valued at $6 billion). According to Nassim Nicholas Taleb, if the maximum loss is $1 and the maximum gain is $1000, the likelihood of failure is not a major concern—as long as one can withstand multiple small losses, there is a potential for significant gains. This reflects the patience of U.S. capital.
5. Blurred Model Boundaries: The U.S. is Undergoing Pressure Tests
The distinction between the Chinese and U.S. models is no longer simply about moving from 0 to 1 versus from 1 to 100:
- China is Moving Towards 0 to 1: China is innovating in open-source models and developing original ADC drugs, creating something from nothing rather than just making expensive products cheaper.
- The U.S. is Also Working on 1 to 100: Reusing Falcon 9 rockets has reduced launch costs to record lows, representing a victory in efficiency.
However, the U.S.'s "high-temperature model" is being challenged by its own policies (such as immigration policies that drive away talent) and financial pressures (such as the $725 billion in capital expenditures). Whether it can continue to innovate or just waste resources remains to be seen.
This article does not favor one side over the other; rather, it uses the metaphor of "two ports of a machine" to illustrate that both chaos and innovation in the U.S. are part of the same system—just as the tents on Market Street and the server racks three blocks away are both contradictory and interdependent. The future competition may not be about who is smarter, but about who can better handle mistakes and adapt quickly.