First, Benedict Evans has a fascinating and detailed analysis of Amazon's financial reports. Read the whole thing.
He shows how Amazon's strategy is not to generate and distribute profits, but to re-invest their cash flow into staring and developing businesses. Starting each business absorbs cash, but as they develop they turn around and start generating cash that can be used to start the next one.
Amazon’s OCF margin has been very roughly stable for a decade, but the FCF has fallen, due to radically increased capex.Here Evans shows capex as a proportion of sales, showing a relentless rise starting in late 2009.
That is, if Amazon was spending the same on capex per dollar of revenue as it was in 2009, it would have kept $3bn more in cash in the last 12 months.
As far as I can see, in the "cloud" business only Google has the same synergy between an established business, and a cloud business. Other competitors don't need the cloud scale of investment to support another, much larger existing business. They have to treat their cloud investments as a stand-alone business, which is much less efficient. And they are much smaller than AWS. So they aren't going to survive. IBM and Microsoft, I'm looking at you.
Second, Chris Mellor at The Register looks at the hype surrounding the "all-flash data center" and makes the point that Dave Anderson of Seagate has been making for years.
That leaves us with the view that all-flash data centres are not feasible at present. They may become feasible if the cost of flash falls to near-parity with nearline and bulk storage disk but there is another problem: the flash foundry capacity to build the stuff just doesn't exist.Not to mention that generating a return on a $7-9B investment requires that the product it builds be in the market for many years. Flash technology is approaching its limits, so the time during which flash will dominate the solid-state storage market with its premium pricing is short, too short to generate the necessary return.
In terms of exabytes of capacity, worldwide disk production is vastly higher than that of flash, and with flash fabs costing $7bn to $9bn apiece it is likely to remain so.
This is no small matter. An all-flash data centre would need approximately the same number of TB of storage as current all-disk or hybrid flash/disk data centres.
The flash foundry operators are paranoid about avoiding loss-making gluts of product, having seen the dire effects of that in the memory industry, with its persistent huge losses and dramatic supplier consolidation. They will be slow to bring new flash fab capacity online.
They are working towards increasing flash capacity by increasing wafer density through cell geometry shrinks, and also through building flash chips with stacked layers of cells, so-called 3D NAND.
These in themselves won't allow the flash industry to take on any substantial portion of worldwide disk capacity in the next few years. That requires many new fabs and there is no sign of that happening.