Down goes the Dow

This from Meta's latest 10q, Note 1:

In January 2025, we completed an assessment of the useful lives of property and equipment, which resulted in an increase in the estimated useful lives of
most servers and network assets to 5.5 years, effective January 1, 2025. Based on the servers and network assets placed in service as of December 31, 2024, the
financial impact of this change in estimate included a reduction in depreciation expense of $2.29 billion and an increase in net income of $1.96 billion, or $0.76
per diluted share, for the nine months ended September 30, 2025.*

Back in August I posted this from Tomas Hirst:

"The other fun part of the story is that capex initially shows up as net
profit in the ramp stage (before depreciation) and, given how
information equipment is treated in macro accounts, as higher real
growth so the story at least *appears* to have tangible support in micro
and macro analyses."

"But to keep that going you have to constantly increase your pace of
capex so that it keep rising faster than the pace of depreciation. Fun
game! Bit vulnerable to, err, when the music stops though."

"There's much funnier option that the game of capex chicken continues
even though it becomes increasingly clear that the moat is dissolving,
in part b/c of ratchet dynamics discussed above & in part due to a
falling but nonzero chance of winner-takes-all"

What's interesting about Burry's chart; MSFT, AMZN, GOOG have large legacy cloud computing platforms ( Q2-2025 revenue respectfully; 29.7B, 19.8B, 12.9B ), so all their capex isn't exclusively AI related. ORCL's revenue was $3B and META is starting from scratch. META, it's all about AI, yet it's depreciation useful life is more ( so are the others ) than the sector leader AMZN.

*This is GAAP accounting.