Amazon, Microsoft, Meta and Alphabet are expected to spend about $725 billion on capital projects in 2026.
Much of that spending is tied to artificial intelligence, cloud computing and data centers. The scale of the investment highlights a planning problem facing large technology companies.
A modern platform may operate digital services, physical infrastructure and a large workforce at the same time. Each part of the business grows differently, but all of them compete for the same money.
Gaurav Walawalkar is a finance manager who has worked on long-range planning for commerce and logistics operations. His work focuses on connecting capital spending, staffing and revenue forecasts instead of treating them as separate budgets.
“A platform does not grow one number at a time, so you cannot plan it one number at a time,” Walawalkar said.
One company can contain several businesses
Traditional financial models often begin with a revenue forecast. The company then estimates how much it will spend on employees, equipment and other needs.
That approach becomes harder when different parts of the business operate on different schedules.
A data center may take years to design and build. A warehouse needs equipment and workers before it can handle more orders. A software product may require years of development before it produces meaningful revenue.
Each investment also affects the others. Building a warehouse creates staffing, technology and transportation costs. Expanding an online service may require more computing capacity, customer support and security spending.
If those connections are missing, every individual budget can look reasonable while the complete plan remains impossible to carry out.
“Most planning gaps come from missing connections, not math errors,” Walawalkar said. “The spreadsheet adds up perfectly and still describes something that cannot happen.”
Capital staffing and revenue need to move together
Walawalkar said an integrated model should show how a change in one part of the company affects the rest.
If a business adds physical capacity, the model should also estimate the employees, technology and operating costs needed to use it. It should then connect those costs with the expected revenue and the time required for that revenue to arrive.
That does not mean every estimate will be correct. Long-term forecasts always involve uncertainty.
The purpose is to make assumptions visible before the company commits money.
A plan should show what happens if construction is delayed, hiring takes longer than expected or customer demand grows more slowly. It should also identify which projects depend on the same resources.
Without that view, two teams may assume they will receive the same funding, workers or technical support.
Scenario planning shows where a plan could break
A single forecast can create a false sense of certainty.
Platform companies often need several scenarios showing different levels of demand, cost and timing. A company might model a base case along with faster-growth and slower-growth outcomes.
The useful question is not only how one number changes. Finance teams also need to understand what that change does to the rest of the plan.
A delay in infrastructure may push back revenue. That revenue shortfall may limit hiring or leave less money for another project. Those effects can continue through several parts of the business.
“The dangerous number is usually the second-order effect,” Walawalkar said. “The number you are staring at is rarely the one that hurts you.”
Scenario planning can help leaders find those connections before they become expensive commitments.
A model must be readable
A financial model is not useful if only its creator understands it.
Operations, engineering and commercial teams all need to know what the plan assumes. They also need to understand what will happen if those assumptions change.
Walawalkar said the logic should be clear enough that leaders from different departments can follow the same chain between investment and expected results.
“A financial model is a communication tool before it is a calculation tool,” he said.
When teams cannot understand the central model, they may create their own versions. The company then ends up with several plans competing for one budget.
A shared model does not require every department to agree. It gives them a common set of assumptions to question and update.
Static annual budgets can fall behind
Annual budgeting remains common, but platform operations can change much faster.
Shipping prices, cloud costs, hiring conditions and customer demand may move throughout the year. A forecast approved months earlier may no longer reflect what the company is experiencing.
In a HackerNoon article about global e-commerce expansion, Walawalkar argues that finance teams should connect their forecasts more closely with operational information.
That could include warehouse use, delivery costs, inventory movement, product returns and customer demand. These signals can reveal problems before they appear in a quarterly financial report.
Rolling forecasts allow a company to update its expectations as conditions change. They do not replace the annual plan, but they can keep it from becoming outdated soon after approval.
Capital allocation becomes part of the strategy
The largest technology platforms are making investments whose financial results may not be clear for years.
That makes capital allocation more than an accounting exercise. Leaders must decide which projects deserve funding, which ones can wait and what evidence would justify changing direction.
An integrated model can make those decisions easier to examine. It can show the expected cost of a project, the resources it requires and the other work that may be delayed as a result.
No model can remove the risk from a large investment. It can make the risk easier to see.
“For a platform, the plan is a map of how every part of the business depends on every other part,” Walawalkar said. “Build that map well and you can move fast and still know where you are going.”
©2026 Cox Media Group
