CAPEX Optimization with AI: StratePlan vs. State-of-the-Art Algorithms
State-of-the-art algorithms optimize individual investment projects. StratePlan optimizes the entire investment portfolio using proprietary mathematical technology.
Companies invest substantial sums annually in new facilities, infrastructure, digitalization, research, and strategic projects. The key challenge lies not only in identifying economically attractive individual projects, but also in selecting the optimal combination from all available investment opportunities.
Qualified software developers can rely on established state-of-the-art algorithms for this purpose. Methods such as mixed-integer linear programming, branch-and-bound, dynamic programming, or GRASP are well-established in the scientific community and available through various optimization libraries.
StratePlan by mAInthink takes technology a step further. Its hybrid optimization architecture specifically combines the strengths of various established algorithms and enhances them with proprietary mathematical methods developed by Dr. Igor Kadoshchuk.
The goal: to maximize the overall economic contribution of an investment portfolio without increasing the available CAPEX budget.
In a generated comparison example, StratePlan achieves 151.5% more economic profit than the IT comparison solution shown—with the same budget cap.
1. CAPEX Optimization: Why Good Individual Projects Don’t Guarantee an Optimal Portfolio
Traditional investment planning often evaluates projects based on individual metrics such as return on investment (ROI), net present value, payback period, or strategic priority.
These metrics are important. However, they do not automatically answer the crucial question:
Which combination of all available investment projects yields the highest overall economic contribution given the existing budget and resource constraints?
A project with a high individual return does not necessarily have to be part of the economically best overall portfolio. A combination of several other projects can achieve a higher absolute contribution to earnings with the same budget.
As the number of potential investment projects increases, the combinatorial decision space grows exponentially.
CAPEX optimization is therefore not merely a matter of project evaluation, but a mathematical selection and allocation problem.
The Limitations of Traditional Investment Planning
- Individual projects are evaluated without systematically considering all possible combinations.
- Budget decisions are often based on rankings, priorities, or historical planning processes.
- Project dependencies and resource constraints complicate the selection process.
- Alternative investment portfolios are often examined only to a limited extent.
- The opportunity costs of a selected project combination often remain unknown.
Mathematical optimization methods can address these challenges by evaluating project combinations based on a defined objective function and binding constraints.
2. State-of-the-art algorithms for CAPEX optimization
There are numerous established mathematical methods for combinatorial investment and portfolio optimization.
A qualified software developer from an in-house IT department can implement these methods or use them via available solver libraries.
Mixed-Integer Linear Programming (MILP)
MILP models investment decisions using integer or binary decision variables, linear objective functions, and linear constraints.
This method is suitable, among other things, for selecting investment projects subject to budget, capacity, and resource constraints.
Branch-and-Bound
Branch-and-Bound systematically explores the decision space and uses mathematical bounds to exclude subspaces that cannot yield a better solution.
This method forms an important foundation for numerous exact optimization algorithms.
Dynamic Programming
Dynamic programming breaks down optimization problems into smaller subproblems and reuses previously computed intermediate results.
It can be particularly effective for structured selection problems, such as variants of the knapsack problem.
GRASP – Greedy Randomized Adaptive Search Procedure
GRASP combines randomized constructive search strategies with local improvement.
The method can generate high-quality approximate solutions for complex combinatorial optimization problems within limited computation times.
Genetic Algorithms
Genetic algorithms use population-based search methods in which candidate solutions are refined through selection, recombination, and mutation.
They are particularly suitable for complex search spaces and nonlinear problem structures.
Local Search Methods and Metaheuristics
Local search methods gradually improve existing solutions through changes within a defined neighborhood.
Advanced approaches include, among others, Tabu Search, Simulated Annealing, and Variable Neighborhood Search.
How effective are these algorithms?
All of the methods mentioned have specific strengths, prerequisites, and limitations. Their actual performance depends, among other things, on the following factors:
- The mathematical structure of the optimization problem
- Number of projects and decision variables
- Type and number of constraints
- Quality of implementation
- Parameterization of the respective method
- Available computing time and hardware
- Quality of the input data used
State-of-the-art algorithms form a powerful foundation for CAPEX optimization. However, simply using an established algorithm does not guarantee that the economically optimal project combination will be found within a given computation time.
3. StratePlan: Hybrid Optimization with Proprietary Mathematics
StratePlan by mAInthink takes a different technological approach than the exclusive implementation of a single standard algorithm.
The optimization architecture combines selected strengths of various established mathematical methods and enhances them with proprietary methods developed by Dr. Igor Kadoshchuk, CTO and scientific developer at mAInthink GmbH.
The Technological Differentiation
StratePlan does not simply use yet another state-of-the-art algorithm, but rather combines different algorithmic strengths within a proprietary optimization architecture.
The proprietary mathematical methods serve to make the combinatorial search efficient and to identify economically valuable project portfolios.
Three Technological Levels
1. Established mathematical optimization methods
Well-known algorithmic approaches provide proven methods for searching for, evaluating, and improving project combinations.
2. Hybrid combination of algorithmic strengths
Selected features of various methods are specifically combined, rather than limiting optimization exclusively to a single algorithm.
3. Proprietary mathematical extensions
Mathematical methods developed by Dr. Igor Kadoshchuk expand the optimization architecture and form an essential component of StratePlan’s technological differentiation.
The result is a specialized decision intelligence technology for the mathematical optimization of strategic investment portfolios.
4. CAPEX Results Comparison: IT Specialist vs. StratePlan
A test example generated from the StratePlan front end demonstrates how significantly different optimization approaches can impact the economic contribution to results.
300 investment projects with a budget of 1 million euros
Both solutions are given the same project proposal and the same budget cap.
The comparison solution, referred to as the “IT specialist,” represents conventional algorithmic CAPEX optimization. StratePlan uses its proprietary hybrid optimization architecture.
| Key metric | IT Specialist | StratePlan |
|---|---|---|
| Available Projects | 300 | 300 |
| CAPEX Budget | €1,000,000 | €1,000,000 |
| Capital Invested | €931,600 | €984,000 |
| Expected revenue | €1,021,100 | €1,209,100 |
| Net profit | €89,500 | €225,100 |
| Profit / Capital Invested | 9.61% | 22.88% |
| Additional profit | Benchmark | +135,600 € |
| Relative profit increase | Benchmark | +151.5% |
151.5% more profit with the same CAPEX budget
The IT comparison solution generates a net profit of 89,500 euros.
StratePlan generates an economic profit of 225,100 euros under the same budget cap.
This corresponds to an additional economic profit contribution of 135,600 euros, or a relative profit increase of 151.5%.
The return on investment (ROI) based on the capital actually invested rises from 9.61% to 22.88%.
The economic benefit results from a different selection and combination of investment projects, not from an increase in the available CAPEX budget.
Methodological Note: The figures are based on a generated test example. The specific algorithm and computation time of the comparison solution are not documented here. The example therefore does not constitute a general performance comparison against all state-of-the-art solvers.
5. 300 Projects: A Decision Space with 2^300 Combinations
In a binary investment decision, each project can either be selected or not selected.
For N projects, there are theoretically 2^N possible combinations.
For 300 projects, this corresponds to:
2^300 ≈ 2.04 × 10^90 possible project combinations.
Budget and resource constraints reduce the number of permissible combinations. Nevertheless, the remaining decision space can be extraordinarily large.
Why a simple project ranking is not sufficient
Selecting the projects with the highest individual returns does not necessarily lead to the highest overall economic contribution.
A portfolio consisting of several projects with moderate individual returns can generate a higher absolute profit than a combination of fewer projects with high individual returns.
Added to this are dependencies, capacity constraints, project durations, and strategic objectives.
The task of mathematical CAPEX optimization is to identify the economically optimal combination of projects based on a defined objective function.
6. In-house IT development or specialized decision intelligence?
A company can develop its own CAPEX optimization solution. Established algorithms, open-source libraries, and commercial solvers are available for this purpose.
However, developing such a solution requires mathematical modeling, implementation, validation, system integration, and ongoing maintenance.
StratePlan provides a specialized optimization platform whose proprietary mathematical architecture was developed specifically for analyzing complex investment portfolios.
| Criterion | In-House IT Optimization | StratePlan |
|---|---|---|
| Algorithms | Established standard methods and solvers | Hybrid optimization architecture |
| Mathematical Methods | Depending on the implementation | Established and proprietary methods |
| Development effort | In-house development | Provided platform |
| Optimization goal | Depends on the defined model | Maximization of the total economic contribution |
| Project dependencies | To be modeled individually | Can be modeled as additional constraints |
| Scenario analysis | To be developed individually | Integrated portfolio and scenario analysis |
| Deployment | Depends on internal architecture | SaaS, API, or on-premises |
The key advantage of a specialized platform is that companies do not first have to develop their own optimization technology to perform mathematical CAPEX analyses.
7. Decision Intelligence: CAPEX Optimization for CFOs, CEOs, and Executive Boards
Strategic investment decisions are among the most economically significant tasks of corporate management.
The key question is not which algorithm is used, but which investment decision will yield the highest overall economic contribution.
Which projects should we finance?
StratePlan analyzes which combination of available projects achieves the highest total economic contribution within the defined budget and resource constraints.
Which projects should we postpone?
The optimization also identifies projects that are not part of the calculated target portfolio. This “anti-portfolio” provides a basis for discussing postponed investments.
How does the portfolio change with a different budget?
Alternative budget constraints can lead to different optimal project combinations. StratePlan enables the calculation and comparison of such scenarios.
What opportunity costs result from our decision?
The economic difference between a planned project combination and a mathematically optimized one can be reported as lost profit contribution.
How could we have optimized past investment years?
Historical project and financial data can be used to calculate alternative investment portfolios retrospectively and compare them with the decisions actually made.
This transforms investment planning from a mere allocation of budget to a mathematically supported decision regarding the economically optimal allocation of capital.
8. Live CAPEX Simulation in the Boardroom
Traditional investment decisions are often supported by presentations, Excel models, and prepared scenarios.
StratePlan expands this process by enabling the immediate mathematical calculation of alternative investment portfolios right during an executive board meeting.
Typical questions during a board meeting
- What happens if we reduce our CAPEX budget by 10%?
- Which projects should be prioritized if we have an additional investment budget?
- Which combination of projects maximizes our total economic contribution?
- How does the result change if certain projects must be implemented?
- What economic impacts result from changes in resource constraints?
- What alternatives exist to our currently planned investment portfolio?
- What is the difference in results between our plan and the calculated portfolio alternative?
The results can serve as the basis for immediate management decisions.
StratePlan transforms static investment presentations into a dynamic, mathematically supported decision-making environment.
9. What data does StratePlan need for CAPEX optimization?
In the simplest case, getting started with mathematical portfolio optimization requires only three pieces of information per investment project.
- Project ID: Unique identifier for the project.
- Investment Expenditures: Required capital outlay.
- Expected revenue: Projected economic return.
Based on this, a simple investment selection problem can be modeled under a given budget constraint.
Advanced Optimization Parameters
For more complex investment decisions, additional information can be taken into account:
- CAPEX and OPEX structures
- Project durations and investment timing
- Project dependencies and exclusion conditions
- Resource and capacity constraints
- Strategic weightings and priorities
- Budget limits for different planning periods
- ESG and impact criteria
The quality of the optimization depends on the quality of the underlying data and the accurate modeling of the business objectives.
10. Scientific Classification: Result Quality, Computation Time, and Optimality Gap
The performance of mathematical optimization methods cannot be assessed solely based on the name of an algorithm.
Four criteria in particular are relevant for a scientifically sound comparison.
1. Quality of the objective function
What is the economic contribution to earnings achieved by the identified project combination?
2. Compliance with Constraints
Does the solution satisfy all budget, resource, and project dependency constraints?
3. Computation time
How long does the algorithm take to find a solution with the desired level of performance?
4. Optimality gap
How large is the proven deviation of the found solution from a mathematical bound or the global optimum?
11. Economic Benefits: Greater Profit Contribution Without Additional CAPEX
For companies, the mathematical performance of an optimization method is particularly relevant when it leads to better economic decisions.
StratePlan aims to achieve the highest possible total economic contribution from a given investment budget.
Greater Economic Profit Contribution
An improved combination of existing investment projects can increase the expected profit of the overall portfolio without changing the budget ceiling.
Transparent opportunity costs
By comparing alternative project combinations, economic differences between a planned and an optimized capital allocation can be made visible.
Improved capital allocation
The available CAPEX budget can be allocated more strategically to those project combinations that best align with the defined economic and strategic objectives.
Mathematically Justifiable Decisions
The results provide a quantitative basis for investment decisions and enable the systematic comparison of various budget and portfolio scenarios.
Up to 60% higher ROI—without additional CAPEX
mAInthink positions StratePlan as having the potential to achieve significant improvements in economic return on investment in suitable use cases.
The actual benefit achievable depends on previous investment planning, available project alternatives, the objective function, and applicable constraints.
The economic potential does not necessarily lie in additional investments, but rather in a better combination of projects already planned.
The statement “up to 60% more ROI” describes a performance claim dependent on the specific application and does not constitute a guaranteed improvement. It should be distinguished from the illustrative comparison example shown, which demonstrates a 151.5% increase in profit.
12. Your CAPEX budget is set. But is your project mix optimal?
Many companies know which investment projects they want to finance. The much more difficult question is whether this selection actually achieves the highest overall economic contribution within the available budget.
StratePlan makes it possible to analyze this question mathematically.
With the StratePlan Online Decision Service, companies can analyze their project portfolios, calculate alternative investment combinations, and compare the economic results.
Get started with your own investment data
To get started easily, all you need is the project ID, investment expenditures, and expected revenues.
Up to 15 projects can be tested for free.
Try StratePlan now and optimize your CAPEX portfolio
Discover the additional financial contribution that a mathematically optimized combination of projects can deliver.
Same projects. Optimal combination. Better results.