The following is a copy and paste from this discussion with Pierre 303:
Let us take an example where I believe that it is clear that there are no reasons for budget. Imagine a company, which has virtually limitless capacity to produce (for example if they have infrastucture which can serve 1 million or 10 million customer by just buying a couple more servers and having a few more support staff). Such a company is exclusively concerned with getting dollars out of their infrastructure (to cover their fixed costs). I believe, then that it is meaningless to say to a sales team that they need to watch their cost. It does not matter what expenses they have as long as they make a profit and are aware of the cash-flow of the whole company. If that sales manager knows that he/she can sell for 10 million by investing an additional 100.000 dollars, then it is a no-brainer, no more senior management needs to consider this, he should be allowed to do it as soon as possible. There, of course, needs to be limit to how much unexperiensed managers can spend in their first year or so.
My question: Do you know any examples of large companies where many departments and functions are managed without traditional cost/revenue budgets, but where departments are measured on exclusively on the profits they deliver? There also needs to be a method to see to that the company does not run into cash-flow problems, by setting some dynamic limits on the amount of liquidity that can be used. Strategic business-units (SBU) is a way to do this, but a company is usually just split in a few SBUs (like General Motors) and I am talking about splitting up in like a hundred units, which are measured on profits.
If no examples are provided, I will ask a separate question on how such a thing could be implemented and the implications it might have.
If I understand your question, you appear to be focusing on access to capital. It's pretty routine in many large businesses, especially where there is capital-intensivity (infrastructure-heavy businesses, for example), to operate structures giving access to capital somewhat independently of Business Unit structures (that is, focusing on projects and incremental profitability rather than lines of business and existing budgets and plans).
Accounting practice (and, I suggest, common sense) works against any idea that a business may drop financial controls altogether. And budgets, plans, forecasts and the rest are well-understood and make sense in the 'business as usual' state - which such projects become as soon as they are approved.
That said, there's definitely been a trend in businesses I've known over the past decade to give far more discretion on expenditure right through the organisation. That's both organisational good sense (most incremental decisions are best taken at the point of need and by the people involved day-to-day) and financial good sense (if you use a rule of thumb estimate that any human intervention costs c. $50, and bear in mind that most such decisions involve small amounts, then it's obvious that good monitoring and macro-level controls will tend to be vastly more efficient in the short term and more effective in the medium term).
One thing I think you're drawing alongside without stating is that the common sense idea that 'cost' can be determined is actually wrong - dramatically so in the case of capital-intensive businesses. And that, though shocking, is absolutely true. It's necessary to take a view of costs, and important to remember that different (sound) methodologies lead to different (coherent) answers.
Let's take an example. We'll use 'lots of servers supporting lots of services.'
Imagine a new opportunity. Here are different - and valid - approaches to understanding cost.
First, we could say that - certainly at the outset - there's no additional infrastructure cost associated with this new service. So infrastructure cost is zero. This is valid.
Second, we could say that the new service will over time call for more infrastructure. We could project this additional cost, and calculate an effective cost per unit. There's speculation in here, but it's also reasonably observable, so we can adjust the calculation periodically. Now we have an infrastructure cost from day one associated with service usage. If we've got our sums right, over time these 'costs' will exactly cover the extra infrastructure costs arising out of the new service, and never the original infrastructure. This is valid.
Third, we could look at existing internal costings used for services provided over this infrastructure, and apply those. This is valid.
Fourth, we could create a mash-up of all the business projections for different services, allocate total cost in proportion to usage (or revenue, or...) and write those numbers in to plans. This is valid.
Fifth, we could avoid any direct cost estimation or calculation and declare that infrastructure cost is, say, 15% of gross revenue. This is valid.
This list is by no means exhaustive. But the answers are different - structurally, numerically and consequentially. This illuminates the point that in complex, capital-intensive businesses, the key management lever is the ability to choose how costs are allocated across lines of business.