Signal
What has actually changed? Establish the observable movement before interpreting it.
Independent analysis of geopolitics, defence, artificial intelligence, robotics and technological power. I build working theses around mechanisms, dependencies, constraints, time horizons and the evidence that would prove the thesis wrong.
Events are the visible layer of strategy. Beneath them sit capability, incentives, dependency, industrial depth, geography, demographics, military constraints, technology, logistics and economic power.
My analysis starts below the event layer. Instead of treating today's alliances, conflicts, institutions and technologies as fixed, I ask which variables are moving, which dependencies are becoming harder to replace and which constraints are likely to bind.
The objective is not certainty. It is to construct falsifiable theses, expose the causal mechanism, assign a horizon, make uncertainty visible and keep testing the thesis against what happens next.
Political statements and diplomatic symbolism matter, but they are rarely sufficient measures of strategic importance. I focus instead on what actors can actually provide, deny, replace, absorb, or impose.
What can the actor actually do? Military capacity, industrial depth, technology, capital, population, resources, logistics, intelligence, geography, and institutional competence.
What does one actor receive from another that cannot be easily replaced? Dependency often reveals more about a relationship than public declarations of friendship.
If a relationship disappeared tomorrow, how difficult would it be to reproduce its strategic value elsewhere? Scarcity creates leverage.
Current power is only one snapshot. Population, technological progress, economic growth, industrialization, and military modernization change the balance over time.
What happens after the obvious consequence? Strategic decisions alter incentives for third parties, create new vulnerabilities, and generate reactions elsewhere in the system.
What prevents the theoretically optimal strategy? Geography, escalation risk, economics, political legitimacy, logistics, reaction time, and technological limitations frequently decide outcomes.
The question is not merely “Who is friendly with whom?” It is “Who materially changes whose ability to exercise power?”
Strategic foresight becomes more disciplined when evidence, inference, probability and consequence are kept separate. I use a four stage chain to stop a compelling story from masquerading as a forecast.
What has actually changed? Establish the observable movement before interpreting it.
Why should that movement create another consequence? State the causal link explicitly.
What becomes more probable if the mechanism continues, weakens or collides with another force?
What should be watched, hedged, built, delayed or tested because the scenario now matters?
Evidence is not inference. Inference is not probability. Probability is not certainty. A good strategic model keeps those boundaries visible.
These are working theses rather than claims of certainty. Each one states a mechanism, a direction of change and a set of forces that can be tracked over time. The point is not to sound certain. The point is to become easier to prove wrong.
The sections below keep the reasoning visible: what is observed, what is inferred, what the thesis predicts, which constraints matter and where uncertainty remains.
The LLM race does not end with better software. Its economically transformative destination may be the convergence of intelligence, agency, memory, and physical embodiment.
The current artificial intelligence competition is commonly described as a race to build increasingly capable language models. I see the model race as an intermediate stage rather than the destination.
Intelligence becomes vastly more economically powerful when it gains persistent agency, multimodal perception, memory, the ability to use tools, and eventually a capable physical body.
Human organizations divide intelligence into occupational boxes because humans are constrained. One individual rarely possesses elite capability across engineering, finance, marketing, sales, law, operations, logistics, and management simultaneously.
Advanced general purpose AI does not necessarily inherit that constraint. A sufficiently capable system could operate across domains because its role is defined by the task presented to it, not by a fixed professional identity.
The same humanoid could be the salesman at 11 AM and the CEO at 8 PM.
Imagine a sufficiently advanced embodied AI system. During the day, it physically travels to customers, demonstrates a product, negotiates, and sells.
Later it returns and analyses the company's financial position, studies advertising performance, creates the next marketing campaign, interviews candidates, modifies software, communicates with suppliers, studies competitors, and recommends strategic decisions.
The important property is not simply that it performs many jobs. It can potentially carry context across all of them. The salesperson, strategist, analyst, engineer, and operator no longer need to be separate information silos.
As general intelligence, autonomous agency, and general purpose robotics converge, organizational scale may become increasingly disconnected from human headcount.
Companies historically scale by adding specialized humans. AI may gradually invert that logic.
In that world, asking how many employees a company has becomes less informative.
The more important question becomes: how much autonomous productive capacity does the company control?
The extreme version is not simply a company with fewer workers. It is a fundamentally different organizational form: a founder, capital, AI infrastructure, and a fleet of digital and embodied intelligences capable of moving fluidly across functions.
Russia's most strategically useful major partner today could become one of the most consequential asymmetries Moscow must manage tomorrow.
Russia and China currently have strong structural reasons to cooperate. China gives Russia access to an enormous neighbouring economy, industrial supply, trade, diplomatic depth, and an alternative pole of power outside the Western system.
But strategic relationships cannot be understood only by asking whether cooperation exists today. The more important long range question is how the relative power of the two partners evolves.
If China's economic, technological, industrial, and conventional military capacity continues expanding relative to Russia's, the relationship may become progressively more asymmetric even if both governments remain politically aligned.
A partner does not need to become an enemy for power asymmetry itself to become strategically uncomfortable.
A dramatically stronger China could possess increasing leverage over markets, infrastructure, investment, supply chains, resource demand, and the strategic options available to Moscow.
The paradox is therefore structural: the relationship that helps Russia reduce dependence on the West can, if insufficiently balanced by other major relationships, create a different form of dependency.
Strategic diversification becomes more valuable as dependence on any single major power increases.
The conventional interpretation is that deeper Russia × China cooperation reduces India's relative importance to Moscow. My long range hypothesis is almost the reverse.
The stronger China becomes relative to Russia, the greater Russia's incentive may become to preserve major relationships that provide strategic alternatives.
India cannot replace China, nor would that be the point. Its value is that it represents something structurally different: another enormous Asian power centre with independent strategic agency.
India's value is not that it can become another China. Its value is precisely that it is not China.
A powerful India gives Russia another major market, another diplomatic relationship, another industrial partner, and another independent centre of Asian power.
This means Russia × China cooperation and Russia India cooperation need not be treated as mutually exclusive. Under increasing Sino Russian asymmetry, maintaining strategic depth through India could become more, not less, rational for Moscow.
Modern strategic relationships increasingly derive value from technology, co development, manufacturing, intelligence, geography, market scale, and shared strategic autonomy rather than formal alliance labels alone.
Aircraft, engines, defence technology, and industrial cooperation are visible. Less visible is the value India provides through market scale, Indo Pacific relevance, industrial opportunity, diplomatic autonomy, and partnership with a major power outside a simple bloc structure.
Defence technology, intelligence, manufacturing, joint development, autonomous systems, sensors, and security cooperation can gradually transform a buyer × seller relationship into a deeper technological ecosystem.
Some of the most consequential future partnerships may be built around co developed technologies, interoperable industrial capacity, and mutually valuable strategic infrastructure rather than treaty language.
Conflict in one theatre can alter the incentives, confidence, capabilities, and risk calculations of actors in another without requiring centralized coordination.
Strategic analysis often makes one of two mistakes. Either every simultaneous event is treated as evidence of hidden coordination, or no relationship is considered possible unless direct coordination can be proven.
There is a third possibility: actors independently react to a changed strategic environment.
In examining India × Pakistan tensions, Afghanistan × Pakistan tensions, and Baloch insurgent activity, I became interested in whether pressure applied in one area can indirectly alter opportunity structures elsewhere.
Forces, attention, readiness, and political bandwidth are redirected toward one theatre.
Other actors observe stretched resources or changing perceptions of vulnerability.
Independent actors reassess the probability that escalation, resistance, or pressure can succeed.
Actors do not need to share a command room to benefit from one another's pressure.
The analytical challenge is therefore to distinguish coordination from correlation while still recognizing genuine system effects. Resource diversion, demonstrated vulnerability, morale, political attention, and perceived opportunity can transmit pressure across theatres.
Nuclear weapons make territorial revision extraordinarily dangerous, but they also create a deeper strategic question: where exactly are the boundaries between deterrence, conventional conflict, and uncontrolled escalation?
My interest here is not in advocating conflict. It is in understanding one of the hardest problems in modern strategy: how conventional military objectives interact with nuclear deterrence.
A serious state cannot assume the strategic environment will remain permanently unchanged. It therefore requires contingency thinking across multiple possible futures rather than one fixed plan.
Any analysis of a geographically limited conventional conflict between nuclear armed states must therefore include much more than a comparison of armies.
The deeper question is whether nuclear deterrence eliminates the possibility of limited conventional objectives or instead creates a narrow and exceptionally dangerous space beneath the nuclear threshold.
The question is not whether advanced combat aircraft remain powerful. It is whether the relationship between cost, survivability, reaction time, autonomy, and destructive capability is changing underneath them.
Modern combat aircraft are among the most sophisticated machines ever built. Future sixth generation platforms may combine stealth, sensor fusion, networking, electronic warfare, autonomous wingmen, and extraordinary information processing capability.
But technological sophistication alone does not answer the strategic question of what a system contributes inside a complete defensive or offensive chain.
If an incoming weapon can complete its mission before an expensive defensive platform can reliably detect, reach, engage, and destroy it, what exactly is the platform's value in that particular chain?
Fighter aircraft do not guarantee interception. Neither do ground based air defence systems. Both alter probabilities.
That distinction becomes increasingly important as weapons become faster, more autonomous, more numerous, and potentially much cheaper than the systems attempting to stop them.
Imagine a high value incoming missile. Detection occurs. Air defence attempts interception and fails. Aircraft must become airborne, establish useful geometry, detect the target, obtain a firing solution, engage, and successfully destroy it before impact.
Every step consumes time and introduces another probability of failure. Increasing the probability of interception can still be enormously valuable, but it should not be confused with guaranteed protection.
Future defence planning may increasingly revolve around probability per dollar , not sophistication per platform.
Autonomous systems introduce another asymmetry. If comparatively inexpensive unmanned systems can threaten platforms costing tens or hundreds of millions of dollars, military economics begins changing even before the expensive platform becomes technically obsolete.
The relevant questions therefore include:
The thesis is not that fighter aircraft become useless. It is that their optimal role may change as the economics and speed of warfare change around them.
AI and robotics may change not merely companies, but the variables through which national power itself is measured.
Industrial power has historically depended heavily on human labour, factories, capital, energy, technology, natural resources, and demographic scale.
Advanced AI and robotics could change the relative weight of those variables.
Consider two hypothetical countries.
One has a population of 500 million but relatively weak autonomous infrastructure. Another has only 80 million people but controls enormous fleets of highly capable digital agents, industrial robots, autonomous logistics systems, and machine operated manufacturing.
Which country has the larger effective labour force?
Which has greater industrial mobilization capacity? Which can produce more military equipment? Which can conduct more research? Which can operate more infrastructure?
AI may not make population irrelevant. It may make population less equivalent to productive capacity.
This has particularly important implications for aging societies, countries experiencing demographic decline, and technologically advanced states with relatively small populations.
Japan, South Korea, Europe, Russia, China, India, and the United States could all experience the geopolitical consequences differently.
If advanced intelligence becomes a foundational productive resource, control over compute, semiconductor supply chains, energy, frontier models, robotics, data infrastructure, and autonomous manufacturing becomes more than an economic issue.
It becomes a component of national sovereignty.
A forecast becomes useful only when its assumptions are visible and when there is a way to determine later whether it was right, wrong, early, or based on a driver that ceased to exist.
Establish the measurable trend before extending it into the future. Separate current evidence from interpretation.
Identify the mechanism linking the observed forces. A thesis should explain why one development could produce another.
Define a horizon, identify confidence, list the major drivers, and state what developments would weaken or invalidate the forecast.
| Forecast / Thesis | Horizon | Primary Drivers | Confidence | Status |
|---|---|---|---|---|
| LLM intelligence converges with general purpose robotics | 10 30+ yrs | Multimodal AI, agents, robotics, compute, model capability | Developing | Tracking |
| Organizational roles increasingly converge under general AI | 10 30+ yrs | Agency, tool use, persistent memory, embodiment | Developing | Tracking |
| China's rise increases India's diversification value to Russia | Long range | Sino Russian asymmetry, Indian scale, strategic autonomy | Moderate | Tracking |
| Air power shifts toward distributed human machine systems | 10 25 yrs | Drones, autonomy, missiles, cost asymmetry, networking | Developing | Tracking |
| AI partially decouples productive capacity from population | 20 50 yrs | Robotics, autonomous industry, compute, energy | Long range | Developing |
The subjects vary. The underlying interest is consistent: how power, technology, incentives, constraints, and strategic behaviour interact.
A compelling story is not enough. What mechanism connects cause and consequence?
Statements matter, but material capability determines what actors can actually sustain.
“Friend,” “ally,” and “partner” reveal less than understanding who depends on whom, for what, and how replaceable that dependency is.
The balance that exists today may not be the balance that matters twenty years from now.
Hypotheses should remain hypotheses until evidence justifies stronger language.
A forecasting record is useful only if failed predictions are retained, examined, and used to improve the model.
My aim is to study the shifts in power, technology, economics, defence, industrial capacity and strategic dependency while they are still weak signals, then keep the reasoning visible as the evidence changes.