AI Content Chat (Beta) logo


AppNexus Infrastructure for a Connected World

Contents • Founders • Why AppNexus? • Differentiation / Competitive Analysis • Timeline • Financials

Founders • Brian O’Kelley, CEO - former CTO of Right Media. Invented, built, sold, and launched the Right Media Exchange. Scaled the RM environment from 10 to 1500 servers and felt the pain of not having AppNexus. • Michiel Nolet, VP of Product - former Director of Analytics at Right Media.

Why AppNexus

Right Media • Scale - MySpace is going to send an challenges extra 1B imps/day of volume... next week • Globalization - Yahoo! Japan wants to do a pilot, but only if servers are in Japan... • Integration - let’s integrate a contextual analysis platform... to all 6 datacenters...

Solving the Problems • Scale: Cloud computing would have been great - flip servers on when we need them • Global: Multiple clouds, with smart routing / load balancing, in key geographies • Integration: Run new services “in-cloud”

Cloud → Nexus • The cloud becomes a “nexus” where applications can consume services • Cloud layer needs to be robust, reliable, easy-to-use, etc to get adoption • Centralized directory, documentation, community, and billing for service providers

Cloud Architecture

Nexuses • One Nexus isn’t enough - applications are moving global quickly • Multiple Nexuses around the world, each with same services and applications • Build tools to help scale applications (data tier especially)

Revenue Sources • Rent high-end 4-core boxes for around $10/core/day (virtualized, of course) • Rent database boxes, storage, bandwidth, backup, and many other things too • Profit-sharing with service providers • Access fees

Competitive Analysis

Differentiators - Near Term • Global (some tough problems to solve there - will be compelling) • Ops tools (very challenging to build a scalable ops layer) • Customer-centric (not exciting, but need to get the little things right - it’s a small world, and trust travels)

Differentiators - Long Term • With network effect (lots of services & lots of applications) will be difficult for a new cloud player to compete - will need to recruit the services to get the apps and will need to recruit the apps to get the services • Lots of tools - idea is to get applications “locked in” to the AppNexus way

Competition • Amazon / EC2 (biggest risk - they have capital, the SOA concept, and momentum) • Rackspace / Gridlayer (hosting companies have the hardware and the customer base, but probably not the SOA mentality) • Google / Yahoo / MSFT etc (not in space now, but have the capex to enter)

Growth Plan

“Ping pong” growth • First applications sign up for the cloud • These clients become leverage to recruit software providers • Co-market with software providers to get more applications on board • Repeat...

Phase 1: The Cloud • Build-out and deploy robust physical and virtual infrastructure to underlie the Nexuses • Can be profitable as a standalone service

Phase 1 Logistics • First location live Jan 2008; 2nd in Mar-Apr • Work closely with initial clients to refine ops layers (monitoring, provisioning, cluster management tools) • Beta launch Jan 2008, Scale to 20 clients by mid- 2008 • Ideal clients are platforms and services (natural SOA consumers and purchasers) • Fund via $2-3MM seed round (plus $2MM of personally guaranteed debt)

Phase 1 Pipeline • Behavioral Marketing (Exelate, Blue Kai, Media Six Degrees) • SaaS (Wide Orbit) • Gaming (Garage Games) • Video CDN (Digital Fountain)

Phase 2: Credibility • Go elephant hunting - recruit the platforms that will make all the services want to be in their Nexus (or vice versa) • Aggressively brand as the high-end of the cloud market - graduate from EC2 Competitor/Cloud-Computing provider • Leverage key differentiators to keep pricing high; continue to build IP

Phase 2 Logistics • Start in Q3 2008 • Raise $10-15MM to ramp team, provide working capital for 2000+ servers • Need global presence - market may be stronger (as is currency) overseas

Phase 3: Nexus • With key clients and critical mass, network effect begins to build • Focus on comarketing with partners - help service providers recruit platforms & apps • Provide arms to both sides of a war that never ends...

Phase 3 Logistics • Begin in early 2010 • Infrastructure side of business should be self-sustaining at this point, so shift to marketing, support, evangelism • Be opportunistic about acquiring facilities • Business should be profitable but raise $25-50MM to expand (and compete)


AppNexus has stepped up its fundraising, bringing in $50 million for its real-time display advertising platform from existing investors as well as Microsoft (s MSFT). The company has now raised $65.5 million over the last three years from investors including Venrock, Kodiak Venture Partners, First Round Capital, Khosla Ventures, Mark Andreessen, Ben Horowitz and Ron Conway.

The company's real-time bidding business counts customers like Microsoft and eBay (s EBAY), as well as eight of the top 15 ad networks. It auctions more than 4 billion ads on a daily basis and is growing fast. While real-time bidding is still a small part of the market, AppNexus said it expects the sector to grow to 20 percent of the display market by 2012.

New York City-based AppNexus, which pitches itself as a scalable cloud service, runs out of three data centers including one co-located at a Equinix facility in Amsterdam to serve Europe and the Middle East.

AppNexus CEO Brian O'Kelley was formerly CTO at Right Media (acquired by Yahoo (s YHOO)), and CTO Mike Nolet was director of analytics there. President Michael Rubenstein was general manager of DoubleClick's ad exchange, then directed the product for Google (s GOOG). Google also recently bought the demand-side platform Invite Media. AppNexus has been discussed as an acquisition target for AOL (s AOL) and Microsoft.

Related content from GigaOM Pro (subscription req'd):

FEATURES/ Cross Border | Jul 24, 2013 | 16110 views

mployees at AppNexus, a bustling advertising technology firm near New York City's peaceful Madison Square Park, don't often interact with the boss. The five-year-old company, with 547 employees, has been growing too fast for that. Plus, its 6-foot-5, 240-pound CEO, Brian O'Kelley, can be, well, a bit irascible. Office folklore has it that he once threw a stapler at a software engineer. (O'Kelley vehemently denies that version, saying it was a tape dispenser thrown on the floor with no employees present.) And he was forced to resign from his previous job for clashing with the boss and being hard to work for.

On a recent Wednesday O'Kelley confessed to Forbes that one of his staffers had earlier that day burst into tears after a project meeting with him. O'Kelley says the tears came from personal issues, "but I was thinking, 'Oh God, I'm sensitive to my reputation.' I know the impact I have on people."

O'Kelley can be overbearing, and friends say he's mellowed a bit, but people put up with him and invest millions in him because he's arguably the most brilliant software engineer in New York's brimming ad-tech community. He's known as the inventor of ad exchanges, the massively interconnected bidding systems that account for a growing portion of sales of web banner and rectangle ads.

AppNexus servers process 16 billion ad buys per day and handled an estimated $700 million in ad spending last year, giving it the biggest reach on the open web after Google.

Yahoo and Facebook run bigger exchanges, but they primarily sell their own inventory. But AppNexus's technology is widely regarded as the fastest and most flexible out there. If there's an exchange that can rival Google in the $13 billion market for web display ads-and everyone in the industry wants alternatives-it's AppNexus.

"I believe there will be a natural duopoly in the ad exchange space instead of eight or nine players," says Bill Wise, CEO of ad infrastructure firm Mediaocean and former president of ad network Right Media. "The two that will emerge are AppNexus and Google."

AppNexus will take in an estimated $130 million in fees this year, up 85 percent from 2012. Ebay, Microsoft and Facebook are all big partners. The company is not yet profitable but has raised $140 million in venture capital to invest in new products and global growth. In January it was valued at $625 million; now it's likely closer to $850 million. After Tumblr it's among the most valuable private firms in New York's Silicon Alley. And it probably would not have happened had it not been for the sizeable chip on O'Kelley's sizeable shoulder that spurred him to prove doubters wrong.

In less than a generation the ad-tech industry has gone from the equivalent of manual placement of banner ads on web pages to continuous auctions in which hundreds of ad-serving decisions are made in the milliseconds before a page loads. The lion's share of banner ads (as opposed to the much larger search-ad market) are bundled into networks that apply the relevant user-targeting data. These groups of networks in turn form exchanges, giant pools of liquidity allowing publishers to sell their inventory faster and buyers to get a fairer price. The primordial ad-serving tech was created by DoubleClick in the mid-1990s, a firm sold for $1.1 billion in 2005 to private equity firm Hellman & Friedman and eventually to Google for $3.1 billion in 2007.

The most valuable descendant of Double-Click to date has been Right Media, founded in 2003 by DoubleClick executive Michael Walrath (and sold to Yahoo in 2007 for $850 million). Right Media commercialised the first big ad network, but started out as a glorified sales desk. Walrath needed a chief technology officer to build his automated ad network, and he found a brash young consultant and programmer named Brian O'Kelley.

O'Kelley built Right Media's network but was soon interested in the newer idea of connecting networks to form exchanges. In 2005 O'Kelley hatched a plan to license a version of Right Media's network to an Israeli software firm called Cydoor so it could bid from within Right Media's own network. "The first time an auction connected across the two, I shouted, 'Eureka!' and no one had any idea what I was talking about," O'Kelley says. (He calls himself "100 percent" the sole inventor of the exchange; others say Right Media's management team had joined in the idea.)

This article appeared in the Forbes India magazine issue of 26 July, 2013

More articles from this writer

Next Article in Cross Border

Like this article? Subscribe to Forbes India

You might also want to read...

Inventing a new way to make money on the Internet, and competing with behemoths like Google didn't come easily to AppNexus founder and CEO Brian O'Kelley.

He had to get fired-more than once.

"I've gotten fired a few times, and every time it was the same reason," O'Kelley told CNBC. "I saw this huge incredible opportunity to do something that would change the world and the people I worked for said, 'hey, that's not realistic.'"

O'Kelley's rise to a billion-dollar valuation came with a few bumps. He was fired as Chief Technology Officer at Right Media, which was automating advertising networks, the day after the company was sold to Yahoo for $850 million in 2007.

But the AppNexus chief is in fairly good company. The late Steve Jobs was once famously fired from Apple, the company he eventually turned into a global brand with his knack for technology innovation.

O'Kelley now works for himself, with some taking to calling him the king of ad technology at one of New York's most valuable start-ups.

"I feel like I have to live every day at work as if it could be my last," O'Kelley said. "I have to take the risks that allow us to potentially fail because it's the only way to learn."