Technology
Wired World: 35 Years of Submarine Cables in One Map
You could be reading this article from nearly anywhere in the world and there’s a good chance it loaded in mere seconds.
Long gone are the days when images would load pixel row by pixel row. Now, even high-quality video is instantly accessible from almost everywhere. How did the internet get so fast? Because it’s moving at the speed of light.
The Information Superhighway
The miracle of modern fiber optics can be traced to a single man, Narinder Singh Kapany. The young physicist was skeptical when his professors asserted that light ‘always travels in a straight line’. His explorations into the behavior of light eventually led to the creation of fiber optics—essentially, beaming light through a thin glass tube.
The next step to using fiber optics as a means of communication was lowering the cable’s attenuation rate. Throughout the 1960-70s, companies made gains in manufacturing, reducing the number of impurities and allowing light to cross great distances without a dramatic decrease in signal intensity.
By the mid-1980s, long distance fiber optic cables had finally reached the feasibility stage.
Crossing the Pond
The first intercontinental fiber optic cable was strung across the floor of the Atlantic Ocean in 1988. The cable—known as TAT-8*—was spearheaded by three companies; AT&T, France Télécom, and British Telecom. The cable was able to carry the equivalent of 40,000 telephone channels, a ten-fold increase over its galvanic predecessor, TAT-7.
Once the kinks of the new cable were worked out, the floodgates were open. During the course of the 1990s, many more cables hit the ocean floor. By the dawn of the new millennium, every populated continent on Earth was connected by fiber optic cables. The physical network of the internet was beginning to take shape.
As today’s video from ESRI shows, the early 2000s saw a boom in undersea cable development, reflecting the uptick in internet usage around globe. In 2001 alone, eight new cables connected North America and Europe.
From 2016-2020, over 100 new cables were laid with an estimated value of $14 billion. Now, even the most remote Polynesian islands have access to high-speed internet thanks to undersea cables.
*TAT-8 does not appear in the video above as it was retired in 2002.
The Shifting Nature of Cable Construction
Even though nearly every corner of the globe is now physically connected, the rate of cable construction is not slowing down.
This is due to the increasing capacity of new cables and our appetite for high-quality video content. New cables are so efficient that the majority of potential capacity along major cable routes will come from cables that are less than five years old.
Traditionally, a consortium of telecom companies or governments would fund cable construction, but tech companies are increasingly funding their own submarine cable networks.
Amazon, Microsoft and Google own close to 65% market share in cloud data storage, so it’s understandable that they’d want to control the physical means of transporting that data as well.
These three companies now own 63,605 miles of submarine cable. While laying cable is a costly endeavor, it’s necessary to meet surging demand—content providers’ share of data transmission skyrocketed from around 8% to nearly 40% over the past decade.
A Bright Future for Dark Fiber
At the same time, more aging cables will be taken offline. Even though signals are no longer traveling through this network of “dark fiber”, it’s still being put to productive use. It turns out that undersea telecom cables make a very effective seismic network, helping researchers study offshore earthquakes and the geologic structures on the ocean floor.
AI
Visualizing the Top U.S. States for AI Jobs
Nearly 800,000 AI jobs were posted in the U.S. throughout 2022. View this graphic to see a breakdown by state.

Visualizing the Top U.S. States for AI Jobs
Much ink has been spilled over fears that artificial intelligence (AI) will eliminate jobs in the economy. While some of those fears may be well-founded, red-hot interest in AI innovation is creating new jobs as well.
This graphic visualizes data from Lightcast, a labor market analytics firm, which shows how many AI-related jobs were posted in each state throughout 2022.
In total there were 795,624 AI jobs posted throughout the year, of which 469,925 (59%) were in the top 10. The full tally is included in the table below.
Rank | State | Number of job postings | % of total |
---|---|---|---|
1 | California | 142,154 | 17.9% |
2 | Texas | 66,624 | 8.4% |
3 | New York | 43,899 | 5.5% |
4 | Massachusetts | 34,603 | 4.3% |
5 | Virginia | 34,221 | 4.3% |
6 | Florida | 33,585 | 4.2% |
7 | Illinois | 31,569 | 4.0% |
8 | Washington | 31,284 | 3.9% |
9 | Georgia | 26,620 | 3.3% |
10 | Michigan | 25,366 | 3.2% |
11 | North Carolina | 23,854 | 3.0% |
12 | New Jersey | 23,447 | 2.9% |
13 | Colorado | 20,421 | 2.6% |
14 | Pennsylvania | 20,397 | 2.6% |
15 | Arizona | 19,514 | 2.5% |
16 | Ohio | 19,208 | 2.4% |
17 | Maryland | 16,769 | 2.1% |
18 | Minnesota | 11,808 | 1.5% |
19 | Tennessee | 11,173 | 1.4% |
20 | Missouri | 10,990 | 1.4% |
21 | Oregon | 10,811 | 1.4% |
22 | Washington, D.C. | 9,606 | 1.2% |
23 | Indiana | 9,247 | 1.2% |
24 | Connecticut | 8,960 | 1.1% |
25 | Wisconsin | 8,879 | 1.1% |
26 | Alabama | 7,866 | 1.0% |
27 | Kansas | 7,683 | 1.0% |
28 | Arkansas | 7,247 | 0.9% |
29 | Utah | 6,885 | 0.9% |
30 | Nevada | 6,813 | 0.9% |
31 | Idaho | 6,109 | 0.8% |
32 | Oklahoma | 5,719 | 0.7% |
33 | Iowa | 5,670 | 0.7% |
34 | South Carolina | 4,928 | 0.6% |
35 | Louisiana | 4,806 | 0.6% |
36 | Kentucky | 4,536 | 0.6% |
37 | Nebraska | 4,032 | 0.5% |
38 | Delaware | 3,503 | 0.4% |
39 | New Mexico | 3,357 | 0.4% |
40 | Rhode Island | 2,965 | 0.4% |
41 | New Hampshire | 2,719 | 0.3% |
42 | Hawaii | 2,550 | 0.3% |
43 | Mississippi | 2,548 | 0.3% |
44 | Maine | 2,227 | 0.3% |
45 | South Dakota | 2,195 | 0.3% |
46 | Vermont | 1,571 | 0.2% |
47 | North Dakota | 1,227 | 0.2% |
48 | Alaska | 970 | 0.1% |
49 | West Virginia | 887 | 0.1% |
50 | Montana | 833 | 0.1% |
51 | Wyoming | 769 | 0.1% |
The following chart adds some context to these numbers. It shows how the percentage of AI job postings in some of the top states has changed since 2010.
We can see that California quickly became the primary destination for AI jobs in the early 2010s, presumably as Silicon Valley companies began developing the technology.
California’s share has since declined, with a significant number of jobs seemingly moving to Texas. In fact, many tech companies are relocating to Texas to avoid California’s relatively higher taxes and cost of living.
The 10 Most In-Demand Specialized Skills
Lightcast also captured the top 10 specialized skills that were required for AI-related jobs. These are listed in the table below.
Skill | Frequency (number of postings) | Frequency (% of postings) |
---|---|---|
Python | 296,662 | 37% |
Computer Science | 260,333 | 33% |
SQL | 185,807 | 23% |
Data Analysis | 159,801 | 20% |
Data Science | 157,855 | 20% |
Amazon Web Services | 155,615 | 19% |
Agile Methodology | 152,965 | 19% |
Automation | 138,791 | 17% |
Java | 133,856 | 17% |
Software Engineering | 133,286 | 17% |
If you’re interested in a career that focuses on AI, becoming proficient in Python is likely to be a good first step.
-
Batteries2 weeks ago
How EV Adoption Will Impact Oil Consumption (2015-2025P)
-
Money4 weeks ago
Visualizing the Assets and Liabilities of U.S. Banks
-
Wealth2 weeks ago
Ranked: The World’s Top 50 Endowment Funds
-
Central Banks3 weeks ago
Visualized: Real Interest Rates by Country
-
United States2 weeks ago
Charting the Rise of America’s Debt Ceiling
-
Money3 weeks ago
Comparing the Speed of Interest Rate Hikes (1988-2023)
-
Misc2 weeks ago
Ranked: The Cities with the Most Skyscrapers in 2023
-
War3 weeks ago
Map Explainer: Sudan