As business school folklore has it, you make more money producing potato chips than by producing memory chips.

The times are making that adage particularly fierce. Food inflation is steep and memory deflation is steeper and, once again, the mathematics of memory seem to bellow the same question: what kind of lunatic wants to own a flash memory factory, let alone plunge billions into building more?

The answer to that question may just be emerging. Last week, Samsung, the South Korean group, expressed interest in buying the Silicon Valley giant SanDisk. It does so at a time when industry inventories are big and the market is flooded with cheap memory. Even with its name at the centre of a possible bid battle, SanDisk’s shares took a near 6 per cent nose dive yesterday because analysts just cannot see when the rot will stop.

So, with global spot prices of flash memory screaming into the abyss, and even the mammoths of the sector admitting they wield virtually no price control, the deal looked strange. It makes only a little more sense when you realise the calculation that Samsung has made: pouring cash into a business with hopeless economics through a $3.5 billion-odd (£2 billion) bid for SanDisk might be shooting itself in the foot, but it would be also shooting its great rival, Toshiba, in a far more vital spot.

Nobody yet knows what would happen to Toshiba’s existing $16 billion joint project with SanDisk to build a colossal new chip plant should the company fall into Korean hands. To judge by Toshiba’s shares — down about 10 per cent since the Samsung/SanDisk tie-up was mooted — the market is assuming that the outcome will be messy, expensive or both.

Samsung is also in a slightly stronger position than Toshiba. Unlike its Japanese rival, for whom land constraints force it to build new-generation factories on space vacated by knocking down previous plants, Samsung simply finds an empty spot somewhere in Korea and leaves the older factory humming away, its construction costs long since depreciated and off the books.

But neither consolidation of a competitor nor punishment of an arch-rival solves the underlying misery of chip making and the sector’s fairly hideous outlook. The real trouble is two-fold. The first is that memory — in flash and DRAM forms — is commoditised at an astonishing pace. Each new generation of factory and chip is bludgeoned with price falls almost immediately: analysts at Deutsche bank, for example, suspect that the market price of 1 gigabyte of DRAM memory is at a piffling $1.50 below the cash cost of producing the thing.

The crisis for flash, meanwhile, is that despite the huge global market for MP3 players, digital cameras and other memory-hungry little gizmos, there has been no “killer application” to drive sales and demand to a point where the producers may seize back some price control.

Looming over the sector, and possibly representing an especially dark cloud for Toshiba and Samsung, has been the much-touted industry push towards solid-state PCs. For some years, the perfection of a usable flash-memory only computer has floated around the electronics industry as a plausible but technologically fiddly dream. On paper, if hard disks were universally replaced by flash memory, that would be the killer application to restore the fortunes of the chip makers.

In reality, though, things might turn out rather differently. Intel, the large American CPU maker, is looking more and more interested in becoming a dominant force in flash memory. If the flash-based PC takes off, say analysts, Intel would surely seek to leverage its global dominance in CPUs and bundle its processors with a generous helping of its own memory. Bad news for Toshiba, bad news for Samsung but, from Intel’s point of view, potentially bad news for the old business school adage.

Times Recommends

• Lies and cover-up at MG Rover exposed by report
• High oil prices push manufacturing costs up
• Primark founder steps down as chief executive