SAN FRANCISCO---As many legacy enterprise tech giants look toward mergers and acquisitions to grow (or at least stop the bleeding), startups could be the best bet in the long-term, based on comments from prominent venture capitalist Vinod Khosla.
"I'm not saying big companies don't have a role," said Khosla while speaking at the Structure cloud summit on Wednesday. "Some of them have done pretty well because they understand. But startups discover where the new markets are."
Khosla pointed back just a few years ago to when Cloudera and Hortonworks started growing in attention, arguing established tech companies weren't paying attention to the Hadoop open source framework, which eventually became buzzworthy in IT.
Partnerships do work in some of these areas, Khosla admitted, positing that there are markets that are stable and there are markets that are high-growth.
"It's not like Amex will stop buying from IBM. They're probably still buying mainframes," Khosla speculated.
As one of the founders of Sun Microsystems, Khosla opined about how "the notion of distributed computing" started at Sun in the 1980s, fostering a new class of architectures and software systems.
Yet, no one innovated upon these systems until the last decade or so, Khosla lamented, highlighting Google, Facebook and Amazon Web Services built their own architectures because the old ones were so inefficient.
Khosla suggested if you look at the old guard such as Hewlett-Packard, Dell and EMC -- setting aside Microsoft for a moment -- these have all been "pretty good businesses, but mostly they spent their time engineering financials."
That set back innovation, Khosla concluded.
Furthermore, none of these companies have built a system for DevOps, Khosla observed, which he argued has become so much more important than it was 10 to 15 years ago.
What we need going forward, he continued, is very different from the old model of the enterprise.
Proof it can be done: Khosla insisted Microsoft has done a reasonable job in committing the time and resources for enterprise innovation.
However, Khosla predicted that most innovation - estimating between 80 and 90 percent of it - will come from startups, reiterating at one point that venture capital figures into this as well.
"Innovation happens because you imagine a future, and make it happen. Not because you read a report," Khosla remarked.
Khosla gushed that he is also very excited by the idea that one might no longer think of a computer as a computer, but rather the data center as the computer.
Looking closer at specific major cloud companies, Khosla observed Amazon is starting to make good margins on AWS while adding later that Microsoft is also doing increasingly well. He also briefly underscored the geographical proximity of the two tech giants' campuses in the Seattle metropolitan area, hinting at an "exchange of employees pretty often."
"But Microsoft also understands what the enterprise needs," Khosla observed. "They have the relationships. They understand the requirements."
As for other growing cloud providers, Khosla speculated it is too early too tell.
"Smart companies are the startups reinventing the way data should be so you can do machine learning and analytics," Khosla insisted, adding nothing "should not be in memory and not distributed," stipulating those as two absolute requirements for any system with substance.