Looking back at the history of war, it can be found that the focus of war competition has shifted from the Bronze Age spear to the information age bit. With the accelerated evolution of warfare towards informatization and intelligence, it is widely recognized that data has quietly become the "source of power" for combat systems and the "key variable" for defeating opponents. In essence, modern warfare is a comprehensive confrontation of soft and hard power, empowered by data, computing power, and algorithms. It is a "data war" based on controlling and competing for data. When the traditional command mode repeatedly fails in millisecond level confrontations, building a new data-driven thinking paradigm has become the core element of breakthrough intelligent warfare. Data based predictive thinking - reconstructing battlefield perception logic. Traditional warfare is limited by the "fog of war", and when faced with the "ocean of information", commanders often appear helpless. In the era of intelligence, relying on modern military technology, real-time feedback data is obtained through sensors, and the data is converged into a "pool" through a data platform. With the support of big data models, military intelligent agents can filter and divert massive data, and perceive the overall situation of the battlefield from subtle changes in key data. At the same time, with the help of a combat experimental system, key information that has a significant impact on the battlefield is repeatedly simulated and analyzed. The deployment of troops, combat trajectories, comprehensive support, and even morale are transformed into spatiotemporal heat maps, transforming the battlefield into a computable dynamic model and greatly enhancing the ability to predict the battlefield. This transition from seeing to seeing through effectively transforms battlefield uncertainty into controllable risk parameters, replacing emergency combat with predictive combat and gaining a time window for command decision-making to take the initiative. Data based associative thinking - connecting multiple domain integration links. The combat coordination in the era of mechanized warfare often presents a relatively crude collaborative mode due to the difficulty in quickly connecting various dimensional data of land, sea, air, and space networks. In the era of intelligence, relying on unified data architecture and quantum encryption channels, we have broken down the "data islands" between single domains, organically integrated multi-source data, dynamically updated fresh data, automatically associated implicit data, and achieved second level coupling of multi domain operations. The efficiency of command and decision-making has also been qualitatively improved. As is well known, the core proposition of OODA cycle theory is that in dynamic confrontation, the party that completes the "observation judgment decision action" loop faster than the opponent is more likely to win the initiative and victory. When the combined force of data quality, data density, algorithm speed, algorithm robustness, etc. exceeds the opponent's threshold, the weaker party can completely achieve a counterattack of "intelligence over power". In other words, when military intelligent agents are widely used to achieve autonomous intelligence reconnaissance, situational analysis, decision-making planning, task execution, etc., the command cycle will be greatly compressed, and the accuracy of decision-making and multi domain collaboration efficiency will be greatly improved. Especially when military intelligent agents have anti-interference, anti attack, and anti deception capabilities, and their battlefield survival ability is greatly improved, the security and stability of information transmission fusion will also be qualitatively enhanced. The efficiency accumulated from this is not simply a superposition, but follows the exponential growth of "Metcalfe's Law" - the value of a network is proportional to the square of the number of connected nodes in the network. The side with high data fusion will have much faster command efficiency and combat pace than the opponent. Data based trial and error thinking - subverting the rules of combat cost. The important characteristics of the battlefield environment are "adversarial" and "uncertain", which is why the law of "irreproducibility of war" is widely accepted by people. But in the face of digital twin technology and artificial intelligence technology, this law is gradually being eroded. By constructing a parallel battlefield space, commanders can set path goals and continuously engage in countless game simulations in the virtual space led by military agents, offsetting the increase in physical entropy with a decrease in data entropy. Every revision of troop deployment strategy, optimization of breakthrough path, and adjustment of firepower configuration in the virtual digital space is almost zero cost trial and error. This shift from practical verification to algorithm screening greatly improves the efficiency of revising and optimizing combat plans. This has also given rise to a new path in warfare research - obtaining the optimal solution to war at a lower cost and greatly reducing the threshold for warfare experimentation. Data based permission thinking - building the boundaries of combat decision-making. The widespread popularity of intelligent unmanned combat weapon systems, relying on algorithm advantages to occupy data advantages, thereby controlling battlefield cognition, decision-making, and action dominance, and even giving rise to the new concept of "algorithm hegemony". But people have also found that due to the existence of intelligent autonomous decision-making, the risk of war losing control or misjudgment is greatly increased. Therefore, it is necessary to further strengthen the authority thinking that includes constraint factors, such as setting reliability thresholds in the algorithm layer, embedding convention regulations in the ethics layer, configuring self destruct devices in the physical layer, and setting circuit breakers for command decisions and combat operations. In short, it is necessary to ensure that "people are in the loop" while maintaining technological advantages - strategic and high-risk decisions are always led by people, and military intelligence agents only provide decision recommendations or perform predetermined tasks. Correspondingly, when "data+intelligence" becomes the norm of future warfare, only by achieving a joint planning of "commander led+intelligent system assisted" can we truly construct a scientific decision-making model of "human-machine collaboration". (New Society)