Is it possible to get fiber-like speeds without all the digging? This week, The Five Nine looks at wireless optical technology, using Google spinout Taara as a case study.
Taara CEO Mahesh Krishnaswamy explains what it means to “solve for the monkey,” why wireless optical tech could make all the difference for data centers, and where the technology could go next. Hint: it involves robots.
Catch the video at top, listen to the audio edition and read our transcript below, or watch this and future episodes on YouTube.
This podcast is written and hosted by Diana Goovaerts. It is edited by Diana Goovaerts and Matt Rickman. Liz Coyne is our executive producer. Special thanks to guest Mahesh Krishnaswamy.
To learn more about the topics in this episode:
Taara pitches wireless optical links as a fast fiber fix for data centers
Taara debuts Lightbridge Pro, touts five nines uptime
Taara CEO: Google moonshots, new chip and where it goes from here
Data center headwinds: Fiber industry faces challenges beyond public pushback
Telcos shoulder big risk building fiber for AI data centers
Data centers will need 92,000 miles of new fiber
Diana Goovaerts: When you're building a wireless network that sends data through the air using lasers, you might expect to spend a lot of time thinking about alignment. You might not expect to spend a lot of time thinking about monkeys. But Mahesh Krishnaswamy says that's exactly what happened during early terrestrial deployments of Taara's wireless optical technology in India.
Monkeys climbed the towers, shook the equipment, and gave engineers a new kind of problem to solve.
Mahesh Krishnaswamy: When we deployed Taara the very first times, which is putting it on towers.
We saw that, you know, especially in India, there were a lot of monkeys on the towers, and they were shaking and moving the towers. And we were not able to keep the link up stable enough. We have a saying called 'solve for the monkey.' We literally were solving for the monkey.
And so as a result, we have one of the most stable, one of the most sophisticated and really capable pointing and tracking systems in the world.
Diana Goovaerts: Yes, the team encountered a literal monkey wrench, and it's a useful reminder that connecting the world isn't just about what works in a lab, it's about what happens when the equipment meets the world outside. I'm Diana Goovaerts, and this is The Five Nine.
Today, we're taking a look at a company that offers fiber-like connectivity without laying fiber and asking the question: can wireless optical links help data centers get connected faster? And where do those links fit in when reliability matters just as much as bandwidth? To get to the bottom of things, I spoke with Taara CEO Mahesh Krishnaswamy about what the technology can do, where it runs into trouble, and the market it's chasing.
To start off with, we need a little bit of backstory here. Mahesh describes Taara as an outgrowth of Google's Project Loon. The idea was to take laser communications developed for links between balloons and bring that technology down to the ground.
Mahesh Krishnaswamy: We call it Moonshot compost, where the same technology can be recycled in alternate ways in order to still solve the same problem or an orthogonal problem. Taara came out of a simple idea, like instead of actually trying to put this on balloons and sending it all around the world, what if we were to bring that same technology but bring it down to the ground and start connecting people and offer fiber-like speeds without having to lay any fiber?
And that's the simple idea that we started out with. And since then, over the last eight years, we have incubated inside Google, and as of last year, we have spun out into an individual company with backing from Google but also other investors to operate as a self-standing company delivering this high-speed connectivity using wireless optical communications.
And as we sit right now, we are live in twenty countries, and we are moving large amounts of data for various enterprises all around the world.
Diana Goovaerts: That's the basic proposition. Instead of digging a route for a fiber cable, Taara is creating a link through the air. But what does that actually look like? Well, Mahesh describes two terminals roughly the size of traffic lights and aimed at one another. Once the terminals lock onto each other, the link carries data using a narrow beam of light.
Mahesh Krishnaswamy: Think of it as fiber optics, but without the physical cables. So it's basically the same light that's inside a fiber optic cable that goes through strands of glass. Suddenly, you just take out the entire medium, and now you're using the same wavelength, same data, but just transmitted over the air.
Ours is a very, very narrow beam, which is the size of a chopstick that is pointing between these two terminals very far apart. And we are able to move vast amount of data, we're talking twenty gigabits per second bidirectionally over large distance of up to twenty kilometers.
Diana Goovaerts: So again, we're talking about narrow beams of light here rather than a floodlight. And Taara concentrates the light between two endpoints rather than using radio frequency spectrum like what's used in cellular networks. He says the lasers are safe for the eyes and the links don't require spectrum licenses.
But this is still a point-to-point connection and it isn't blanket wireless coverage, and that distinction matters.
For data centers, the pitch isn't just another way to move bits. It's another way to get connected while waiting for fiber. Mahesh pointed to permitting, construction and shortages of fiber components as obstacles that are slowing data centers down, and having costly GPUs sitting around waiting for connectivity is a problem.
That's where Taara sees its niche.
Mahesh Krishnaswamy: One of the key things that we are really trying to solve for is the time to deployment and the time to market. So while you're waiting for your permits or approvals and stuff like that, what if you can get a lifeline? What if you can get connectivity to the construction site or even the early versions of the server, so you can actually get them up and running?
There's no value in having bought a whole lot of compute, but it's just sitting, waiting for the data centers to come online.
Diana Goovaerts: Installation is part of the appeal. Mahesh says an app and a built-in camera will help with pointing, and a link can be brought up in a matter of hours with a very small amount of training. But Taara's tech isn't just about providing a temporary fix. Mahesh said it can also be used as a redundant link that isn't vulnerable to cable cuts.
Mahesh Krishnaswamy: Even afterwards also, after the data centers are up and running, a lot of enterprises need to cross-connect to them. We also need multiple redundant paths of connectivity. So eventually, even when fiber comes in, you can leave this in as a redundant circuit where you have a different failure mode compared to fiber. There's no fiber that needs to be cut here.
Diana Goovaerts: So this isn't necessarily a choice between fiber and lasers. It's about the right technology at the right time and having a backup in case one fails. Of course, one 20 gigabit link doesn't answer every data center bandwidth requirement, but Mahesh said multiple links can be combined and pointed to a recent test.
Mahesh Krishnaswamy: We did recently a test of a hundred and sixty gigabits per second by having multiple of these things just, you know, a few meters apart, and you can stack multiple of these things on a single pole.
And as long as you're pointed roughly in that same direction, and they're separated by a few meters, you should be able to have this narrow chopstick beam being able to transmit a combined data rate of more than a hundred gigabits per second.
Diana Goovaerts: The CEO obviously wouldn't disclose their full roadmap to us, but he did say that the company is working towards higher data rates. But if the promise is speed and flexibility, the counterweight to all of this is availability, and Mahesh was upfront about one particular weakness.
Mahesh Krishnaswamy: So our Achilles heel is actually fog because we are transmitting using light, so anything that disperses light tends to be a challenge. And so fog is typically the biggest culprit. And so the way we work around that is a couple of ways. First of all, we can keep the distances short, so we are able to kind of navigate through those.
But the second thing that we can do is we can back it down to another alternate frequency or another fiber if it is aerial also, which is unreliable. But the combination of those two will be able to provide a very high reliability. And then the last part is we can also reroute just like fiber through multiple paths, so we can create a mesh using these wireless optical networks, so that way you're also overcoming the specific fog in one particular localized area.
Diana Goovaerts: If you're weighing something like this technology for a critical connection, the questions are concrete. How does the proposed link perform in local weather? What capacity remains when it switches to backup? And do alternate paths actually avoid the same problem? And there's another requirement that no clever bandwidth figure gets around: The terminals need a clear view of one another.
Mahesh Krishnaswamy: One of the biggest challenges - I wouldn't call it a really a challenge but a constraint - is basically to have clear line of sight between the two points. Oftentimes there's something or other. There's a new building that's coming up, or there's a construction site, or there is a tree that just grows. So the key thing is to make sure you have clear line of sight between the two in order to be able to get access to it.
The good news is that unlike radio frequencies where you have to be absolutely a very big, large antennas that needs to go on towers, we are very versatile.
Diana Goovaerts: Mahesh said the equipment can go on rooftops, towers, poles, tripods, but flexible mounting doesn't eliminate the need for access to those sites or a clear optical path that stays clear. So avoiding a trench doesn't necessarily mean avoiding every deployment constraint.
Beyond data centers, Mahesh said he sees a use case for Tara in serving things like robots, drones and autonomous vehicles. Specifically, he talked about moving large amounts of data off of those machines and uploading stuff like training data and model weights.
So you can think of a charging stop for an autonomous vehicle as an opportunity to exchange data. Mahesh said next-generation products using a photonics chip could make the equipment smaller and lighter, but making it work would also require the machines and the places they visit to be equipped with those optical connections
Mahesh Krishnaswamy: I think as we go further, machine-to-machine communications is gonna consume vast majority of the traffic that's gonna be on the internet, and they don't get tired. So how do you move all of that data? And I think the only way I can think of is optical-based networks. They have a lot of data.
You're talking terabytes of data that needs to be off-boarded, but they also need constant weights and training data that needs to be uploaded to those machines. You cannot attach fiber to each one of these autonomous vehicles and robots and drag them around, so you have to do this in a wireless way.
So we are seeing a very strong pull from these markets, which are physical AI, robotics, autonomous vehicles, drones, where they are looking for a lightweight solution which can offload large amounts of data. Think of it like Wi-Fi or 5G, but on steroids.
Diana Goovaerts: He added it can also be used for broadband use cases where laying fiber is tricky. Think things like river and railroad crossings.
Mahesh Krishnaswamy: Physical terrains like river crossings, railway crossings, mountains, forests, all of these things are really easy for us to cross and traverse through without having to dig through even a single meter of fiber inside those. What we did in Africa between Brazzaville and Kinshasa. They're separated by the Congo River, which is one of the deepest and fastest flowing rivers, and they were not able to run fiber between those two points, and the only alternate was to run five hundred miles of fiber through a smaller, lower rate area and just within a matter of days, we've been able to bridge that connection. It's about four and a half, five kilometers that we've been able to bridge connectivity.
Diana Goovaerts: So where does that leave wireless optical? Well, perhaps the strongest use case is a specific one, a connection that needs to be available before fiber is ready, an additional path once fiber is installed, or a high-capacity exchange at a place where the endpoints can see each other. The harder questions are also specific.
Local weather, site access, clear sight lines, aggregate capacity and what the backup can actually deliver. And the monkey story brings us back around to that same lesson. The technology is only useful if it can handle the conditions it encounters outside the lab. Lasers don't make those questions disappear. They give network builders another way to answer them.
That's all the time we have for today. Make sure you like and subscribe wherever you are listening or watching, and stay tuned to fierce-network.com during the week for news and updates. We will catch you next time!