Is real depreciation expense more than reported?

I was reading Warren Buffett's latest annual report and there was a part where he talked about his railway company, BNSF. He gives a detailed explanation of why BNSF didn't meet his profit expectations:



That got me thinking, is a similar factor at play here with McMahon? MAH similarly has large amounts of expensive equipment that must be routinely replaced. If, say for example, a Caterpillar truck with a useful life of 10 years that cost $10m 10 years ago, is depreciated at $1m a year, and now needs to be replaced in 2024, at a cost of $15m, then surely that depreciation expense didn't, after all, accurately measure the true expense incurred by its owners?

I've always been perplexed at how MAH pays so little dividends relative to its reported NPAT. The net profit that's reinvested into the business never seem to generate good returns. The logic for paying out more of its NPAT as dividends instead of reinvesting back into the business seemed like a no brainer. Have I been wrong now, and that reinvestment of majority of NPAT back into the business is in fact non-discretionary and sustaining capex?

(note: I ask this now more out of curiosity as I bought this earlier at 12.5c and sold into the subsequent rise. But I have always wondered why management didn't pay higher dividends.)