Monday, June 2, 2014

Can Google Achieve a Satellite Internet Cost Breakthrough?

Google plans to spend more than $1 billion on a fleet of new satellites to provide Internet access to underserved regions and people, launching perhaps 180 new satellites, likely using a middle earth network architecture.

What is not clear is the revenue model. Google might be envisioning a wholesale revenue model, which means it must find “on the ground” retail partners. Though a retail model using the same infrastructure is conceivable, that would be a challenging business model.

Gree Wyler, founder of satellite-communications startup O3b Networks, is running the project for Google. O3b's former chief technology officer also has joined Google. That might suggest a wholesale model, necessitating access partners, is at least a part of the effort.

To the extent that middle mile facilities are a key impediment in many emerging Internet markets, the new satellite capability could provide a solution for backhaul to the core of the Internet.

As with any network, there are trade-offs. Generally speaking, satellite will provide less bandwidth than optical facilities, but with the upside of faster deployment at lower capital investment.

Of course, a satellite solution can combine backhaul and access, serving end users directly, or providing backhaul to retail providers. It isn’t yet clear what Google will want to do with its new fleet of satellites.  

Separately, Google also is trying to buy Skybox Imaging, which would provide both satellite and command capabilities, in addition to additional new mapping capabilities, arguably at lower cost per satellite.

Google also bought Titan Aerospace, a manufacturer of drones, and also has been investigating balloon-delivered Internet access.

As has been the case with Google’s earlier investments in municipal Wi-Fi and free airport Wi-Fi, investments in mobile spectrum and even O3b itself, Google investigates numerous access methods, and might eventually commercialize some of them.

Google will provide Wi-Fi access at Starbucks locations, replacing AT&T, and also has commercialized Google Fiber.

Whether Google will succeed at hitting its hoped-for cost targets with the new satellite initiative is unclear. But skeptics doubted Google could turn Google Fiber into a commercial venture as well.

The issue, as always, is whether Google can find some new breakthrough on the capital investment, operating cost or business model fronts, compared to all others who have investigated or who use the same MEO architecture.

To be sure, the cost of supplying satellite bandwidth arguably has fallen over the past couple of decades. By some estimates the latest generations of satellites will feature capacity about 300 percent higher than current generation satellites, and perhaps 33 percent lower cost.

Those with long memories also will recall earlier end-user-focused efforts to create low earth orbit or middle earth networks in the past. To be sure, costs have gotten lower over the past few decades, with developments both in launch costs and the satellites themselves.

Google will of course also have to secure rights to use satellite spectrum.

Some will argue that the key innovations will not come in the areas of bandwidth or cost per megabyte of delivered capacity, but instead in business model innovations. Still, Google and others are creating new competition for legacy providers of Internet access, including now even O3b, one of the newest contenders.

Google will gain by relying on “new technology.” But some will question the magnitude of performance and cost advantages to be gained by using the latest platforms.

As with Google Fiber, with used both traditional suppliers of access and transport gear as well as some home-grown customer premises equipment, the trick will be to innovate in the operations model.

O3b launched an initial fleet of about a dozen satellites, investing about $1.3 billion in that effort. Google plans to launch an order of magnitude more satellites than did O3b.

Even assuming some cost advantages, scale alone will be a cost issue. Even with savings of perhaps 30 percent, Google plans a fleet that is 10 times bigger than what O3b deployed.

How much bandwidth per customer such an approach can yield always is an issue with satellite delivered services. Retail price per gigabyte also can be an issue.

And then there are the business model issues. O3b provides both “middle mile” wholesale transport to third parties and direct to end users. That means some amount of channel conflict is inherent in the business model. Google might face the same dilemma.

How much does O3b cost compared to other available access options, after O3b backhaul is priced in? It is not easy to say. Google will face the same fundamental pricing issues.

O3b itself only says the “real answer depends” on the partner business model and assets. That will be true for Google as well, even if Google will attempt to innovate on the operating cost model, as it has with Google Fiber.

O3b argues it will “compete strongly where high bandwidth, low latency services are required such as IP trunking and mobile backhaul,” as part of a total end user delivery system. Obviously, much depends on the cost profile of the actual access services partner.

In other words, O3b is an efficient supplier of backhaul. It is not yet so clear how efficient O3b will be in an “end-user direct” mode. Google will face the same fundamental concerns.

But Google likely will try to trim its capital costs by using different suppliers than O3b, including using “captive” or “in-house” supply where possible.

O3b, on the other hand, is using existing suppliers of earth station equipment and conventional launch methods for its middle earth orbit (MEO) approach, and therefore is more limited in terms of cost containment.

O3b relies on several established satellite equipment manufacturers, including
Viasat for gateway and enterprise customer earth stations (7.3m and 4.5m antennas).

For “smaller customers” (antennas of 1.8m and 2.4m, probably most suitable for a wholesale customer), General Dynamics earth stations will be used with a range of modems and hub management systems from Comtech, ViaSat and Gilat.

O3b uses 1.1-meter and 2.2-meter earth statiions for its maritime service, produced by Orbit.

Transmission systems are provided by CPI, Comtech, Norsat and NJRC.

O3b has has said US$1.3 billion in initial financing covers the cost of building and launching the first 12 satellites and running the business up to full commercial deployment. Google of course plans a fleet 10 times bigger.

Marketing and operating costs would mount from there, especially if the business leans towards retail, rather than wholesale.

Google will try to make choices that lower breakeven thresholds. It will use Kymeta Corp. antennas that have no moving parts and are controlled by software, which reduces manufacturing and maintenance costs.

Some argue that could reduce costs substantially. If Google Fiber provides any useful guidance, Google might be able to achieve a cost breakthrough.

No comments:

Many Winners and Losers from Generative AI

Perhaps there is no contradiction between low historical total factor annual productivity gains and high expected generative artificial inte...