Big Data is in a serious pickle. On the one hand, the smaller data analytics and processing companies, while enjoying growing revenues, are struggling with razor thin margins and just don’t have the economies of scale to achieve escape velocity into consistent profitability. On the other hand, the big recognized names in data processing like IBM (NYSE: IBM), Hewlett Packard Enterprise (NYSE: HPE), and Cisco Systems (NASDAQ: CSCO) are all struggling. All three have seriously underperformed the Nasdaq over the last decade (counting HPE as part of HP), and recovery looks far off.

The silver lining is that given Big Data is growing at an exponential rate (over 90% of the data ever created was created since 2011), the solution to these growth problems can seemingly be found there. Indeed, Big Data is IBM’s fastest growing segment, spurring $26 billion of investment, $17 billion of which was spent on 30 acquisitions in total. A full third of IBM Research’s spending is on Big Data.

It’s a similar story with Cisco and HPE. Cisco’s Data Center has been its fastest growing major product category for the last five years. The historic HPE split from HP earlier this month was itself an attempt to isolate the faster growing data services sector from HP’s (NYSE: HPQ) bread and butter PC and printer business.

Given that the most recognized names in Big Data are struggling or even shrinking overall but growing in that Big Data segment, and that the smaller companies are growing but do not have the economies of scale that the larger firms enjoy, it seems inevitable that we will see further roll-up in this sector over the next few years. Here are 3 possible acquisition prospects for IBM, HPE, and Cisco.

Datalink

A look at Datalink Corporation (NASDAQ: DTLK) is a perfect illustration of what is going on with the smaller companies in the Big Data space. Revenues have grown 28% since 2012 and 2015 looks on pace to grow 20% more since last year. And yet, Datalink’s bottom line has been struggling mightily, and that has taken a major toll on its share price in 2015. Year to date, Datalink is down over 40%.

Datalink has made some notable acquisitions in the past, most recently Bear Data late last year, but these have not spurred sufficient growth as hoped. The Bear acquisition made Cisco Datalink’s largest OEM, which could mean that Cisco is the best choice for suitor here since there is already considerable overlap. A takeover by Cisco would mean lower cost of revenue and higher margins on Datalink’s revenue stream, and likely lower SG & A expenses as operations could be streamlined.

Sysorex

Much like Datalink, Sysorex Global Holdings Corp. (NASDAQ: SYRX) has a rapidly growing top line, but as it stands, margins are too small to maintain profitability. Revenues have grown remarkably from just over $4 million in 2012 to over $63 million by the close of 2014, with growing gross margins as well.

Sysorex was founded by an IBM employee and what is notable about it is the speed and accuracy of its technology. Sysorex’s security product billed AirPatrol has the ability to pinpoint the location of wireless devices to within 10 feet, accuracy which is unmatched by any competitor. Its other main product, LightMiner, is an in-memory off-hard-drive data storage solution that is cache-free and runs 100x faster than competing systems.

With both growing revenues and top of the line data processing technology, Sysorex could be an attractive acquisition for any of the three Big Data companies. While Sysorex has no earnings yet, annual revenues are four times its entire market cap. Sysorex could be acquired very cheaply at these rates for anyone who believes its tech could give a leg up on the competition.

CSP Inc.

CSP Inc. (NASDAQ: CSPI) is another data solutions company treading water. Very generous to its shareholder though with a 7.5% dividend, it is not giving up despite razor thin margins and steep competition from the bigger players. That generosity may be the company’s saving grace as the stock began to rise from the pit once dividends were announced in 2012.

Like Sysorex though, CSP struggles with a very low valuation of only $20 million compared with its high annual revenues of over 4 times that. Most data companies in this tier are struggling with the same issue – revenues are there, but growth in earnings is not. CSP competes directly with many of the companies who manufacture its third-party products, and these include IBM, HPE, and Cisco. While the competition is great for customers, it is making the sector almost unprofitable, as CSP looks set to record a net loss this year.

If this is going to change for any of the smaller companies, and any of the big players that have been struggling over the last decade, operations will have to be streamlined in order for margins to grow. Cost of revenue needs to come down and administrative expenses need to be slashed. On the other side, HPE, IBM, and Cisco all need a shot in the arm to show the markets they are still capable of growth. In the case of Big Data, it seems that roll-up could be a viable solution, and better now while valuatioms are low before the amount of data being created doubles or triples again in the next few years.