Amazon does not require government money to fill Earth’s lower orbit with hundreds of satellites providing high-speed internet, but it might use some help avoiding space trash and allocating bandwidth effectively.
The tech behemoth is rapidly attempting to become a space behemoth, aiming to send thousands of satellites into orbit to establish a network that will deliver high-speed broadband around the world — directly competing with Elon Musk’s Starlink. This entails coping with both the harsh realities of space and the severe realities of space policy, which is closely controlled and frequently funded by Washington, D.C.
David Limp, an Amazon senior vice president who handles gadgets and services such as the Project Kuiper satellite programme, is one of those driving the initiative. Limp sat down with AWN for a wide-ranging interview on the project and its policy needs, discussing the government’s role in opening space to private firms, some of the challenges Amazon confronts in space, and who Amazon sees as its most important potential customers.
Amazon plans to launch its internet service late next year and have half of its 3,200 satellite network operational by mid-2026.
While Amazon claims it does not require — but is willing to accept — government assistance, Washington policymakers are seeking for ways to help the fledgling industry as a whole grow into a new vehicle for American innovation. But, US policymakers are still arguing how best to support the business, which is beset by a slew of regulatory issues. Legislators and federal officials are currently debating how to speed the licencing process for such satellite systems, while agencies argue when broadband subsidy subsidies should be provided.
All of this is intertwined with how to compete with China as it extends into low-Earth orbit.
Although the possibility of space-based high-speed internet excites many because of its potential to connect remote parts of the world, Amazon and SpaceX confront challenges such as coping with orbital trash and keeping the skies pristine for astronomers on Earth. Amazon’s current plans for reducing space junk from its own satellites were recently approved by the Federal Communications Commission, paving the way for these launches.
Limp spoke with AWN about how the government can help the industry, how artificial intelligence may benefit Amazon’s broadband programme, and who the company’s first prospective customers are likely to be.
Amazon has stated that it plans to begin delivering broadband service by the end of 2024. Who could be the first customers?
We begin with a sparser constellation with fewer satellites and southern and northern latitudes — consider Montana, North Dakota, Minnesota, and Argentina and Australia. Our first service will most likely be in those latitudes. My expectation is that we’ll be able to turn things on with hundreds of satellites in 2024, and that the early clients will most likely be enterprise customers to begin with. Consider someone who has a pipeline that they want to connect to an offshore oil well.
Amazon is proposing three types of customer terminals, one of which is a “basic” device that costs $400 or less to manufacture. How much does it cost to produce the other two versions, the larger high-bandwidth and ultra-compact?
The larger antenna is more expensive because it is intended for enterprise use. It has a gigabit performance and more Merlin chips in it. Something like that would be used by Verizon for a distant 5G tower. In terms of construction and materials, the smaller one is far less expensive.
What are the technical restrictions to increasing broadband speeds over time? The US government has supported fibre broadband networks because they are “future-proof” investments.
We attempted to find a sweet spot here at 400 megabits. Will there be some customers who want more? Without a doubt. Yet, with that much bandwidth, you can have many real-time 4K feeds in a home and still have enough of room for telemedicine.
It’s difficult to dispute that it doesn’t match today’s and, at the very least, the foreseeable future broadband needs of customers.
If I’m a congressperson from West Virginia, North Dakota, or Montana, I represent communities where fibre will never be installed. That would be irresponsible for taxpayers because excavating such a long trench is insane. I’d rather give one of those satellite antennae to a customer. Our infrastructure funds should be used to install fibre in cities and possibly dense suburbs. Nevertheless, once you go out into the sparse suburbs and rural villages, low Earth orbit satellites make a lot of sense.
The United States is spending tens of billions of dollars to fund the construction of broadband infrastructure in unserved and underserved areas of the country. Would Amazon be interested in bidding on such grants?
That is not the foundation of our business model. But, if we can assist, and public monies can assist people who are underserved, we would be grateful. Yet we don’t need it to run a successful business. We developed Kuiper to be a stand-alone firm that does not require government assistance, but I commend governments for attempting to accelerate this process.
And do you have any specific requests for the government regarding Kuiper?
Space security. Space is vast, yet it is becoming increasingly congested. We aim to keep space secure and limit the amount of junk in orbit. We will continue to be vocal in support of that agenda.
The second area in which we have been vocal is the responsible use of spectrum. By the way, that applies to us just as much as it does to anyone else. Give it back if they aren’t using it. If they are using it, agree to collaborate with others so that you are not interfering, based on where they are in the order of when their licence was applied.
We’ve been extremely delighted with what we’ve seen around the world, particularly in the United States with the FCC, in terms of their readiness to listen to it.
SpaceX is a major player in this space, having already launched its Starlink satellite service. What competitive advantages does Amazon stand to gain from launching Kuiper?
I don’t think there will be a single winner in this sporting event. There will be several winners here. I don’t believe there will be a large number of constellations – they’re simply too expensive. Over time, there will most likely be a couple commercial constellations and a couple nation-state constellations.
Having said that, I believe there will be far more consumer, industry, and government demand than the limited number of constellations that we will be able to deliver.
We believe that Kuiper can be distinguished from other constellations. We wouldn’t do it if we didn’t have to. First and foremost, customer service – you can call Amazon 24 hours a day, seven days a week, 365 days a year and speak with a live person about where your box is or why your Kindle isn’t working. It isn’t as common in the networking sector.
We run a massive network in AWS every day, and we believe we can apply that knowledge to this.
Finally, there is the AWS connection. We run a lot of these workloads on the cloud, and if you run your streaming media service on the same cloud as Kuiper, you’ll have reduced latency and higher dependability because those things can be more tightly coupled in terms of network and regionality.
What role could artificial intelligence play? Recent AI advancements have gotten a lot of attention in Washington.
I’ve been repeating for a decade that we are living in the golden age of artificial intelligence. The fact that we have reached a new level of AI is a continuation of this golden period.
With the rise of these generative AI algorithms, you’re seeing another step function shift, most notably by huge language models. They, too, are accomplishing incredible things. It’s also visible in DALL-E and ChatGPT. We also employ these within Amazon – Alexa’s teaching model, which is a tens of billion parameter model, huge language model, has essentially taken the majority of the labour out of switching languages. We had to do a lot of work when we created an Alexa function in German and then had to move it to Spanish: tagging and language and linguists had to be engaged, and computer science had to get involved. Now, this enormous language model has figured out how to create a generic language – it can translate all of these various things side by side.
People will be performing all sorts of things via the network in relation to something like Kuiper, the application layer. When you’re operating massive volumes of traffic across a limited resource, sophisticated network management will come in handy. And every time you figure out how to better route that traffic or compress that traffic and make decisions based on that, you provide a better consumer experience. And there is no doubt that the most recent generation of AI techniques will be beneficial.
Washington politicians frequently frame these arguments as a global tech competition between the United States and China, with the goal of cultivating U.S.-based winners. What are your thoughts on that lens?
There is a tremendous opportunity for the public and private sectors to collaborate more closely. AWS has numerous different groups with which it collaborates with the US government. We do not exclusively run data centres for the US government.
Truthful information has never been more important as a gateway to freedom. Moreover, in a world that is far more intricate than it was perhaps 10-15 years ago, there are instruments that can be used to keep the truth hidden. One of these tools is the ability to block connectivity. As a result, Kuiper has the ability to assist the public sector in ensuring connectivity.
What are the most difficult obstacles to taking Kuiper forward? You’ve already mentioned supply chain constraints.
We have resolved the great majority of our supply chain difficulties. If you’d asked me that question a year or 18 months ago, I’d have given you a very different response because we’re living in a very different era.
Space is difficult. Because space is a harsh environment – radiation, heat, and the risk of orbit — there is a lot to learn. We’re up to the task. But it’s similar to when Amazon chose to enter the consumer electronics market and develop its first Kindle. The muscle had to be developed.
We’re developing a similar muscle around space, but we don’t yet know everything there is to know.