If the future of big data is Hadoop, those peddling it still have a long journey ahead of the them.
Newly IPO'd Hortonworks' inaugural results show a company whose losses are growing as its business expands.
Further, it’s not subscriptions to Hortonworks’ implementation of the open-source Hadoop that are growing most – it is professional services.
That translates as Hadoop remaining complicated to install and use for those customers who are finally ready to make the jump and start crunching big data.
That’s an inconvenient truth for a tech firm trying to become the next Red Hat or Salesforce – living off open source software funded by accelerating and continuous subscriptions.
Hortonworks reported a net loss of $90m (£58m), and total revenue that grew 55 per cent to $12.7m (£8m) for three months to 31 December.
For the quarter, Hortonworks saw near equal income from support and professional services – $6.4m (£4.1m) and $6.3m (£4m).
But it was professional services that grew fastest – up 117 per cent compared to 21 per cent for support.
Based upon past performance, it’s professional services that looks to be growing the most in the future.
For the year, Hortonworks' net loss landed at $177.3m (£114m), up from $63.2m (£41m), on total revenue that nearly doubled for the year – up 91 per cent to $46m (£30m).
Again, professional services outgunned support – up 137 per cent to $20.5m (£13.2m) compared to 65 per cent to £25.5m (£16.7m).
Sales and marketing and research and development were separately Hortonworks’ two biggest costs.
The firm predicted revenue will increase 70 per cent in the first quarter and 63 per cent for the full year.
Hortonworks – which floated in December short of its anticipated valuation – claimed to have 332 subscribers on support, adding 99 in the fourth quarter.
Rival Cloudera last week claimed total revenue of $100m (£64.5m) and 525 enterprise subscriptions customers for its fiscal 2015 year.
Cloudera is not public, so its numbers are not verified and are selectively presented.
What the results seem to demonstrate is that paying Hadoop users are relatively small in number and they are paying more for help and hand-holding than use.
The firms will likely argue the latter will come once the hurdle of the form has been overcome and the customers in question are hooked up.
If so, it suggests a big and – for Hadoop hopers – costly initial upfront heft. ®