Sega, and it's not even close.
Sega opened the 1990's with the Genesis/MegaDrive gaining steam before going through the roof with the release of Sonic in 1991, along with some other great titles that actually helped them gain an edge on Nintendo for a while. The Sega Saturn was a mixed bag, and DreamCast came too little, too late, but was a pretty good system, none the less.
Contrast that with Atari, the company that both started it all, and then became the best point of reference for what NOT to do if you ran a video game company.
Atari had a chance to get back into the fight in spring of 1984 with the 7800 Pro System, actually, after the utter catastrophe that the 5200 Super System had been. Problem was, while Warner had planned a mega launch for the Pro System (the sort of launch the Super System should have gotten), they were negotiating the sale of the company to Jack Tramiel at that point, so everything got put on hold...and then, after the sale was completed in early July of that year, the Pro System remained in limbo (and titles in development remained in a holding pattern) while Tramiel and Warner argued over who was going to pay the independent contractor that designed the Pro System for Atari. Tramiel insisted that his purchase of Atari covered the payment, but Warner, it seems, stiffed the developer, then tried to hang Tramiel with the bill for the Pro System.
It would be nearly TWO YEARS before the matter was settled, and even then, Tramiel would find a way to completely screw the whole thing up:
Smart Jack: Rolls out the 7800 Pro System in May of 1986 (a full FOUR months before Sega's NA Master System launch, and Nintendo's full U.S. release of the NES in September of that year), with an $80 price tag- $120 cheaper than either the NES or Master System at launch. (And $170 cheaper than the deluxe release of the NES that included the utterly useless ROB peripheral.)
Jack Ass: Spends more money promoting (and continuing to develop software for) his $50, outdated, 2600, severely undercutting his brand new machine THAT COULD PLAY 2600 GAMES WITHOUT ANY NEED FOR A PERIPHERAL ADAPTER, THUS RENDERING THE 2600 COMPLETELY OBSOLETE in the first place.
That, and he treated retailers like they were competition, rather than the people he needed to sell his product to the public. (Always bad for business.)
Jack Ass: Okay, so, he gets his ass handed to him by both Nintendo and Sega in '86, even with all of those advantages going in:
SERIOUSLY lower retail price than his competitors. (For a better machine than it's given credit for being...mainly because so few developers were given any incentive to develop software for the Pro System, hence nobody really opened up the new machine, pressing the envelope of the tech to it's outer most capabilities.)
The 2600 Legacy: Millions of people with large libraries of 2600 games that would run on the new machine, hence, a massive market of ready-made customers who, not wanting to literally throw away all that money they sank into their 2600 cartridge libraries, would gladly buy a brand new machine that could run that library, so long as new and more advanced games were delivered going forward.
HOWEVER...
Jack's still got something neither Nintendo, nor Sega has the capability of doing in 1987: Releasing a 16 bit console.
Atari had the ST chipset, and, contrary to what a lot of people think, the ST actually sold fairly well in the early going, especially among the creative market. (The built-in Midi was a big selling point for musicians, and it's desktop publishing software, at the time, was better than anything available for Mac or even PC. It sold for a price that wouldn't bust out aspiring game creators, and could have been one arm in a reciprocating cash machine; ST's spike in sales for aspiring developers, more third party support for the ST derived console, driving sales of the console, which, in turn, drives sales of the ST. Trust me, if I actually diagramed out how this actually would have worked, you'd be asleep in under two minutes.)
It would have been the simplest thing in the world for Atari to put the ST tech into a game box, promote it as 'The Future', shift ALL development and marketing resources to the new console (while still supporting the ST, but slowly moving away from the home computer market, with a target of total withdrawal by the end of the decade), and having a very good chance at regaining market share.
INSTEAD...
Jack Ass: Takes antiquated Atari 8-bit home computer line and launches yet another 8-bit console, only this one had ZERO compatibility with the 2600 (although it had...um, dodgy compatibility with Atari 8-bit computer software, and required the purchase of a $200 disk drive to run the full 8-bit computer line of games...well, the ones that would run on the crap box...), ZERO compatibility with the 7800 (not that it mattered, as so little was done with the machine...), and sold about as well as you would expect: It was DOA.
Not one to learn from past mistakes...
Jack Ass: Rather than spending 1988 developing a 16-bit console- his LAST, BEST CHANCE to bring Atari back from the Abyss -Atari wastes it's capital expanding it's computer line into markets it couldn't even hope to penetrate, let alone compete in, and, for video games (the thing Atari was built on), developing a poorly supported handheld console for release in 1989.
Lather, rinse, repeat...
Jack Ass: Splits resources between two separate console projects (both of which would have been pushed as 32 bit consoles, despite not actually being 32 bit consoles), then goes forward with the more expensive and complex system, over the cheaper of the two...because the developers he contracted to develop the Jaguar told him they were further ahead in their development than they actually were...and he believed them), and, by the time Atari is ready to launch the overly complex, more expensive (and, as many would see it, a blatant case of false advertising), Sega and Nintendo have had their own 16 bit systems on the market (and succeeding) for four and two years respectively.
And THAT is how a company goes from industry watershed to death.
Warner Communications poor handling of the evolving market in 1982/83 inflicted the first life threatening (and ultimately fatal) wounds in the company, but that's another story...