One month ago, I wrote a detailed analysis of CoreWeave (CRWV) and showed you why Jeff Bezos shorted the data-center build-out firm.

Well, sorta.

In his 2004 shareholder letter, the Amazon founder described a business much like CoreWeave and demonstrated why businesses with heavy capital investments always fail if they’re not earning their cost of capital. In these situations, revenue growth simply leads to a bigger failure.

Today I want to examine Venture Global (VG), which is in the process of building out one of America’s largest ever capital projects – a series of enormous liquefied natural gas (“LNG”) export facilities that have already cost almost $50 billion.

The question for investors is: is Venture Global a Bezos “teleport machine,” that like CoreWeave is doomed to fail?

Bezos's teleportation machine turned $160 million of equipment into $100 million of annual revenue. His machine was generating 62 cents on the dollar of its initial cost. But it was hopeless because the machine was scrap in four years. Lifetime revenue: $2.48 for every dollar of capital.

As I showed in the August 31 Journal, CoreWeave is much worse. It only gets 16 cents on the dollar out of assets. And they’re worthless even faster: new Nvidia (NVDA) architecture every 12 months.

Venture Global turns $53.9 billion of plant into $17.0 billion of trailing revenue – 32 cents on the dollar. Better than CoreWeave, but worse than Bezos’ teleport machine, which was doomed to fail.

But there’s a critical difference: a natural-gas liquefaction plant does not become scrap after four years like Bezos’ teleport machine. Venture Global charged $941 million of depreciation last year against $47.3 billion of property and equipment, a schedule that implies a useful life of 30 years – 7.5x the teleport machine.