-
An extraordinary wave of new technology is coming to the cloud
-
New mindsets and business models will be needed to build the cloud diaspora
-
Long-term success requires massive investment in R&D
-
2024 is Year Zero for cloud infrastructure
Change is nothing new to the communications industry. For a century and a half, new technologies have arrived and disrupted the status quo – creating opportunities for some and catastrophe for others.
The unprecedented challenge now is in the pacing of these arrivals.
It took over four decades between the invention of the telegraph (1835) and the arrival of the telephone (1876), then almost a century until the first data network rocked up (1969). After that, the cadence increased … a new wireless G standard every 10 years, a major network transition (IP, optical, cloud) roughly every two decades.
Which brings us to the here and now – a decade in which our industry is about to be subsumed by a tidal wave of innovation of unparalleled size, diversity and speed. Going forward, it will be years, not decades, between disruptive new technologies that will blow the doors off entrenched business models.
There are a few reasons for this acceleration — the metamorphosis of the industry from hardware platforms to software everything; the upsurge in Asia comm tech; the realization of how effing big the global comms opportunity really is —but they’re less important than understanding what’s coming down the comms pike and how it will change everything.
And the fact is, as of today, we’re not ready.
Tech iceberg, right ahead!
As a reader of Silverlinings you already know about cloud and artificial intelligence (AI) and are actively working out how to pragmatically deploy both for profit and efficiency in your business (right?).
But following just behind these two game-changers is a fusillade of other technologies, some of which are only now making the transition from science-fiction to science-fact.
Some of these developments are astounding.
When I first started covering free-space optics as a cub reporter in 1989, the products could transmit a 1 Mbps stream of data across a distance of a couple of city blocks, provided it wasn’t raining.
In 2016 the German Aerospace Center demonstrated 1.72 Tbps across a distance of 6 miles – 1.7 million times faster (yes, that’s a million).
The image above provides a non-exhaustive list. From wetware to probabilistic constellation shaping to nanomagnets, it’s a wonderfully diverse assortment. But each of these technologies has something in common: they’re designed to take a specific characteristic of communications networks and improve it by increasing network capacity/speed, making processing more powerful, allowing more data to be stored, or decreasing the size and energy usage of hardware.
As such, none of these technologies are designed to replace cloud or AI, per se. Rather, they are designed to complement them — becoming key components of the world’s cloud communications infrastructure, which will itself be managed using AI.
Not all of these technologies will succeed in going mainstream, obviously, and that’s ok because there’s a lot of redundancy between what they are trying to achieve.
For cloud architects what’s more important than picking a winner is determining what sort of cloud these upgrades will enable us to build and being ready for it. And that, as they say, is where things get really interesting.
D Days
The next generation of cloud networks will be defined by three characteristics (I’ve managed to get them all to start with “D,” because I felt like it).
Distance independence
Today, international data travels via a chain of fiber networks owned by different service providers. It’s a complicated, expensive and inefficient system. In the near future (this decade), we will start to send data up into space, where it will travel around the planet at light speed via constellations of satellites equipped with free-space optics transceivers, before returning to its destination on earth.
Far-fetched? Not at all. There are three companies developing orbital optical networks: AWS, Starlink and Laser Light Communications. (Disclosure: I work for Laser Light, which is the only one of the three companies planning to use free space optical connections down to the planet, rather than radio).
These space networks will deliver the same benefits to communications that space planes will deliver to travel, fundamentally resetting our understanding of how long data takes to get from A to B on (and around) planet Earth.
Data equality
Space optical networks will eliminate the world’s East/West and North/South digital divides, allowing every nation in the world to participate at full speed in the global digital economy.
Imagine the effect of dropping 200 Gpbs of capacity into Kigali in landlocked Rwanda, which currently has to send cross-border traffic to undersea fiber optic cables Kenya and Tanzania. Now extrapolate that by the other 48 countries in Sub-Saharan Africa. Game changing, life changing, world changing.
Data diaspora
Data centers are the area of today’s cloud market set for the greatest disruption in the next couple of decades.
Exponential advances in data processing and storage technology (smaller, faster) eliminate the need to collocate servers and storage in massive data center facilities.
Instead, expect a Great Migration of compute and storage out into the field, or the house, or the pocket, or the pinkie ring, where it forms a vast data diaspora.
Data centers won’t disappear altogether, but they’ll be fewer in number, and they’re going to relocate. One destination is, again, space, where the near-maximum-zero temperatures qualify satellite data centers as the perfect place to house quantum computers. Others will move to the ocean floor — also nice and cold — where older information that is needed less often will live (remember hierarchical storage management? Its time has come, only three decades after it was first proposed!).
There are no hoomans in either location, by the way. Frigid, vacuum-sealed and run by robots – that’s the future of the data centers that survive the move to a data diaspora.
The time paradox
There’s a time paradox at the heart of today’s innovation cycle. As the interval of time between new and game-changing technologies arriving decreases, the length of time into the future that potential users and vendors of the technology need to plan for increases or they run the risk of building a network that’s out of date before it’s even lit up.
Are service providers and vendors doing this now? Not nearly enough.
In the U.S., most of the good work on advanced tech in North America is being done in academia (kudos, nerds).
What’s needed is a group of industry players that have vast experience in espying promising new technologies, commercializing them, and making the tech available by licensing the patents to everyone (even their competitors).
In other words, telecom vendors.
Build to the future
Building for the future is a problem for incumbent vendors, who are used to focusing on the next financial quarter or next financial year.
Instead, they need to develop toward the communications world 10 years or 20 years from now, and building and monetizing it before their competitors do.
This requires vast investment in R&D, something that is incompatible with their current business models.
Cisco spent around $7.6 billion on research and development in 2023, or 13.33% of its $57 billion revenues. It’s also 65% profitable. To be a competitive force in the coming cloud world, the company needs to sacrifice huge profits for deeper investment in R&D, and it could certainly afford to do that. But it won’t, because it’s not really a go-getter any longer (it left its culture of disruptive innovation next to Route 101 sometime back in the last decade) and is now entirely beholden to its shareholders and Wall Street.
Meanwhile, Ericsson and Nokia both spent about 17% of their revenue on R&D in 2022, or $4.6 billion and $4.9 billion respectively — numbers which are unlikely to go up much given their current financial situation.a Still, they spent more than Deutsche Telekom, which says it spent $33 million on R&D on revenues of $125 billion (or 0.02%.... actual lol at German chancers!).
Currently, the only global company that gotten its head around the R&D imperative is Huawei, which spend 25% of revenues in 2022, or about $25 billion and yet another reason why Huawei is giving its Western competitors conniptions.
The best way for vendors and cloud architects to approach the challenge of new technologies is to look at 2024 as year zero for cloud infrastructure… Blow up the old thinking and bring in the new. Or not. The Chinese are already doing this, and if that’s cool with you, why, just sit back and enjoy the new cloud world.
Read more of Steve Saunders' op-eds here.