The evidence is a bit mushy right now. Teradata's fourth quarter was better than expected with an outlook that was conservative. Teradata's unified architecture plugs into Hadoop, an open source platform to analyze big data, well, but some ETL (extract, transform and load) workloads are moving away from the company.
Teradata reported fourth quarter earnings of $112 million, or 68 cents a share, on revenue of $769 million, up 4 percent from a year ago. Non-GAAP earnings in the fourth quarter were 88 cents a share. Wall Street was expecting earnings of 85 cents a share. So revenue was in line and earnings were better than expected. Following a rough third quarter, Teradata can count the results as a win.
As for the outlook, Teradata said its revenue would be up 3 percent to 7 percent in 2014 with non GAAP earnings of $2.85 a share to $3 a share. Those results fell short of the $3.04 a share Wall Street was expecting.
What's unclear is whether Teradata's business will be hurt by Hadoop and by how much. Teradata has incorporated Hadoop into its architecture and melded it with its Aster platform.
Teradata's plan is to meld Hadoop with its data warehouse wares. The plan makes sense since most large customers will have a hybrid approach for the foreseeable future. But like a software company moving from licensing and support to the a business model that revolves around the cloud and subscriptions, the transition has to be managed well.
Approximately one-third of our top 50 customers in the Americas have Hadoop in production now, with the other two-thirds in various stages of evaluations. The spending patterns on Teradata data warehouses for the customers that have Hadoop in production compared to those who did not was similar. We work closely with these customers, and nearly half of them have adopted our Unified Data Architecture. The total number of customers who have implemented our UDA has more than tripled, and we have several more with proof of concepts underway. In addition, our 2013 Aster and Hadoop related revenue was close to quadruple from what it was in 2012. We embrace Hadoop as a key part of our UDA, because we believe that, net-net, Big Data and Hadoop is a benefit for both our customers and for Teradata.
In comments later, Koehler added that the actual revenue hit from Hadoop paled relative to uncertainty among customers.
For the most part, the customers have Hadoop in production, like we said on the last call. The activity is around ETL and moving some of the ETL workload off of their Teradata EDWs, which we're in agreement with and which we said on the last call. So the quantifiable impact of customers doing that was relatively small. Now, if you look at things going forward and the impact on Hadoop, I will refer back again to what we said on the last call. And that is, basically, we did a thorough analysis of our larger customers and basically what we saw is they averaged 20% to 40% of their workloads being done -- being used for ETL. And of the 20% to 40% of those workloads being done with ETL, we think that 20% is a good candidate to be done with Hadoop. So going forward, we will see more workload being moved as it relates to ETL, and we see that as the biggest impact.
Perhaps the bigger risk to Teradata is that the capital spending among its largest customers isn't growing. And that reality will mean more customers will look to Hadoop, which will probably be able to handle more workloads in the future as the open source projects develop. If enterprises decide that they can navigate big data and analytics without big data warehouses and the integrated appliances, Teradata will have more issues than Hadoop. Some analysts think enterprises are already rethinking data warehouse strategies.
Cowen analyst Peter Goldmacher noted:
Teradata said the biggest problem in 2013 was the company’s inability to get its top 50 customers to spend. Management attributes this dynamic primarily to a slowdown in spend, aka, the macro, while our work indicates that it’s these kinds of companies that are most aggressively adopting Hadoop. We interpret the company’s full year results to indicate that customers that are spending with Teradata are more interested in the company’s less profitable mid-range, lower end and lower cost products that carry smaller profit margin. We also see a refreshed go to market strategy from Cloudera, a Hadoop startup that has Teradata squarely in its sights, around positioning its offerings as a Teradata replacement solution at a fraction of the cost. We see the threat from Cloudera and other Hadoop distros increasing over time as functionality and usability are improving dramatically and rapidly to the point where in a few years time, Teradata won’t have any technical competitive advantage. We are seeing early signs that enterprises are rethinking their entire data management architecture and we see very little room for legacy players in this new paradigm.