Thursday, May 30, 2013

Global IP Traffic Will Grow 23% Annually to 2017

Global IP traffic has increased more than fourfold in the past five years, and will increase three fold over the next five years, according to the Cisco Visual Networking Index.  Growth would surprise absolutely nobody. But the relatively restrained rate of annual growth might.

Overall, IP traffic will grow at a compound annual growth rate (CAGR) of 23 percent from 2012 to 2017. Given rates of growth that had been higher in the past, that relative slowdown in growth rates is noteworthy.

Among other notable changes, metro traffic will surpass long-haul traffic in 2014, and will account for 58 percent of total IP traffic by 2017
Metro traffic will grow nearly twice as fast as long-haul traffic from 2012 to 2017 in large part because of the increasingly significant role of content delivery networks, which bypass long-haul links and deliver traffic to metro and regional backbones.
Content Delivery Networks (CDNs) also will carry 51 percent of all Internet traffic in 2017 globally, up from 34 percent in 2012. 
Also, nearly half of all IP traffic will originate with non-PC devices by 2017
In 2012, only 26 percent of consumer IP traffic originated with non-PC devices, but by 2017 the non-PC share of consumer IP traffic will grow to 49 percent. 
PC-originated traffic will grow at a CAGR of 14 percent, while TVs, tablets, mobile phones, and machine-to-machine (M2M) modules will have traffic growth rates of 24 percent, 104 percent, 79 percent, and 82 percent, respectively.
Traffic from wireless and mobile devices will exceed traffic from wired devices by 2016. By 2017, wired devices will account for 45 percent of IP traffic, while Wi-Fi and mobile devices will account for 55 percent of IP traffic. 

In 2012, wired devices accounted for the majority of IP traffic at 59 percent. 

Tablets Replace PCs in Dakar Cybercafe

go 730x276 Google replaces computers with tablets in a Senegal cybercaféThe Equinox cybercafe, located in Dakar’s Medina neighborhood, has replaced PCs with 15 tablets. Available at the same price as a regular cybercafe computer session at about $0.60 an hour, the move illustrates the potential role tablets will play in getting computing devices into the hands of users in many developing regions, according to the Google Europe Blog.

Aside from lower capital and probably operating costs for the cybercafe, use of tablets also consumes less electricity. Smart phones will play much the same role.

Now if we can just get the cost of infrastructure down, in a disruptive way. 




Will Mobile and Fixed Internet Access Business See Substitution?

In 2012, perhaps one percent of U.S. households stopped paying for home Internet subscriptions and relied on wireless access instead, according to Leichtman Research Group. 

That small amount of wireless substitution in the Internet access market will not bother providers of mobile broadband service. 


Video cord cutting, though more prevalent, still amounts to only about a few percent of households.

Zero TV households number about five million, or about three percent of U.S. households.

Arguably, most "mobile broadband substitution" takes the form of users dropping their fixed wireless subscriptions and relying on free Wi-Fi and their mobile broadband plans. 



Some lighter users, who mostly use social networking apps and sites, might find the approach works. People who don't mind moving to places where there is public Wi-Fi might add some amount of watching video. But multi-user households and frequent video watchers will find mobile broadband substitution a difficult proposition.

Wednesday, May 29, 2013

KPCB Internet Trends 2013

Here is Mary Meeker's latest KPCB Internet Trends 2013 presentation. Among the concepts is the labeling of the current era of computing as the "mobile Internet," with the next era something Meeker calls "wearable/everything computing." Some of use might say "pervasive" computing, but its the same idea. 

Computing in each era has moved from "glass rooms" to desktops, then laps, then pockets. In the next era, computing will be embedded in small devices that people wear. Obviously, the use of computing devices as sensors will grow dramatically. 




Smart Phone Adoption Growing at 42% Annually

The smart phone growth rate globally stands at 42 percent annually,  according to the Kleiner Perkins Caufield Byers Internet Trends analysis. 



98% of U.S. Smart Phone Users Do Not Watch Mobile Video


A recent study by Experian Marketing Services suggests nearly 98 percent of U.S. smart phone users do not watch any video on a typical day.


Perhaps you are as surprised as I was to read that statistic, given all the statistics it would be easy to find suggesting that mobile video is growing really fast. The Experian data suggests high rates of growth coexist with low rates of usage because the growth of mobile video consumption is coming from a very low base of users and usage.

To be sure, other studies might suggest that mobile video consumption already is higher than found by the Experian study. Flurry, for example, found that mobile video consumption was nearly eight minutes a day, in March 2012.

Compare that to the two percent of users who do report using video, for five minutes a day, spread over 4.2 sessions, as the Experian study found.

One might suppose that tablets are a more frequent platform for video consumption. Indeed, a  separate study by Ooyala in the third quarter of 2012 suggested mobile video consumption represented about 2.2 percent of total end user video consumption, with tablets at 3.2 percent of total consumption.

That study suggests tablets are a more frequent screen used for consuming video, than the smart phone.

But a study Flurry suggested just the opposite trend, namely that smart phones were generating more video viewing than tablets.

Older data actually tends to support the Experian study’s finding that mobile video usage remains quite low.

In 2011, 91 percent of mobile device users  did not watch video on a particular day, using their mobile device. But other studies suggest average consumption average consumption of nearly 10 minutes a day on mobile devices.


The point would seem to be that the impact of mobile video consumption is yet to be seen on most mobile networks.

Tuesday, May 28, 2013

BYOD Does Not Seem to Have Harmed the Enterprise Phone System Market

At least so far, the "bring your own device" trend in business does not seem to have affected enterprise demand for business phone systems, though pricing seems to be under pressure. 

The global enterprise PBX market (TDM, hybrid, and pure PBXs) represented $1.8 billion in sales in first quarter of 2013, down nine percent from the previous quarter, and down 10 percent from the first quarter of 2012, according to Infonetics Research. 

Price pressure seems to be the big problem, though, since demand in North America and Asia seems either flat or slightly up, in terms of shipments. “The big squeeze is coming from hyper-competitive price pressure all over, with average revenue per line down across the board,” said Diane Myers, Infonetics Research principal analyst. 

Analysts at Dell’Oro Group think it is possible that generally flat revenue trends in the broader enterprise voice markets could change over the next few years, though. “Our view is that the calculation for enterprises has changed, thanks to BYOD and a younger generation of employees comfortable with social media and communicating without a desk phone,” Dell'Oro Group researchers say. 

 “As a result of this shift, many premise-based customers may adopt “Voice as a SaaS” solution in a hosted/cloud infrastructure in the near future,” Dell’Oro Group says. 

Separately, a study sponsored by Cisco suggests that bring your own device policies already are embraced by 89 percent of organizations, principally meaning those organizations allow their workers to bring their own devices for work. 

The Cisco financial analysis was conducted across six countries--Brazil, China, Germany, India, the United Kingdom, and the United States--and suggests BYOD employees are gaining a global average of 37 minutes of productive time per week. 

The time savings range from 81 minutes per week in the United States to just four minutes per week in Germany, where BYOD is viewed more skeptically. 

On average, the current median level of BYOD implementation—what Cisco calls “Basic BYOD”—generates $350 of value each year per mobile employee. 

The number of BYOD devices in the countries surveyed expected to more than double, from 198 million in 2013 to 405 million by 2016, Cisco says. 

Smart phones are the overwhelming choice of BYOD employees, who own an average of 1.7 devices for work and have paid $965 in out-of-pocket costs for them.

Monday, May 27, 2013

Indoor Wi-Fi Vexes U.K. Enterprise Managers


About 39 percent of U.K. large business managers with information technology  decision making responsibility say poor indoor mobile coverage or capacity are problems they have faced.

That is hardly surprising, given the important role Wi-Fi networks already have assumed in supporting mobile device Internet access.

The survey also reveals that 35 percent of IT managers say that they would be prepared to move to a wireless carrier that could guarantee a better indoor solution.

Some 47 percent of IT managers are interested in mobile device management as an operator-hosted service to manage, monitor, secure and support mobile devices in the enterprise.  

About 40 percent of  IT managers also had interest in Wi-Fi as a service from their operator.





What Would it Take for Google Fiber to Attract Investors Formerly Funding Cable and Telco Networks?

A reasonable argument can be made that the actual primary and intended purpose of Google Fiber is to cause other major ISPs to vastly increase their own investment in access networks.

But there could be other eventual outcomes. Google Fiber might become something else, namely a sustainable business. You might argue that Google would not want the huge distraction of creating a huge new ISP business.

But it might not have to. Google Fiber could conceivably become so successful that it could attract the same sorts of investors that traditionally have invested in telco or cable TV networks.

Google Fiber potential hinges on penetration rates, one might argue. In other words, if Google Fiber does only as well as the best traditional “overbuilders,” getting possibly 20 percent of homes as customers, it might be a sustainable business, but just barely.

But if Google Fiber were to reach much higher levels of adoption, up to perhaps 50 percent, it would have financial prospects vastly better than any other overbuilder has achieved.

Google Fiber's core network will cost between $674 and $500 per passing, the former representing Kansas City, Kan. costs, the latter Kansas City, Mo. costs.

It will cost Google $464 to actually connect an Internet access customer, and $794 to connect a customer buying both video and Internet access. Those figures are roughly in line with what other telcos might expect to invest in a similar market.

The key variable is degree of customer penetration, not so much cost of infrastructure.

Assume it costs $600 to build the network passing every location, and then about $450 to connect each actual customer.

At 20 percent penetration, the cost per customer is $450 plus the value of the capital invested to build of the rest of the network that is not serving actual customers. At 20 percent penetration, that means the cost of the network per customer is $3,000. So the cost of serving each customer is $3,450.

For simplicity, ignore the potential value of customers who opt for the “free” access option, and the potential value to Google Fiber if each of those “free” nodes includes a public Wi-Fi element.

At 50 percent penetration, the cost per customer (including both network and customer premises capital) is $1650. At 50 percent adoption, revenue is more than twice as high as at 20 percent penetration and capital investment per customer is nearly 50 percent lower.

Were Google Fiber to reach 50 percent penetration, profit margin might also improve by as much as 100 percent over the 20-percent penetration level.

That might make Google Fiber something that can be replicated on a wide scale, attracting the same sorts of investors that previously might have funded telco and cable TV networks.

Sunday, May 26, 2013

Internet Speeds Have Grown 10X Every 5 Years Since 1990

Access speeds in the U.S. and other markets have grown by an order of magnitude (10x) every five years, since about 1990. If you assume that speeds were about 10 Mbps in 2010, that suggests U.S. consumers will be buying 100 Mbps services by about 2015, and could be buying 1 Gbps services by 2020.

That will be a shock for most ISPs, but would simply follow the pattern of typical bandwidth since 1990.

And that is just what some Internet access speed forecasts would predict. As a rough rule of thumb, speeds have increased by about an order of magnitude every five years.



Availability of 100 Mbps services grew even more: 448 percent between 2010 and 2012. Availability of 50 Mbps services grew 160 percent between 2010 and 2012, the
NTIA study found. 

Australian access speeds show the same trend, with access speeds doubling about every five years.




In August 2000, only 4.4 percent of U.S. households had a home broadband connection, while  41.5 percent of households had dial-up access. At that time, the effective price for a 1-Mbps connection might have been $234.

A decade later, dial-up subscribers declined to 2.8 percent of households in 2010, 68.2 percent of households subscribed to broadband service, with effective prices per Mbps of perhaps a couple to a few dollars.

The availability of 100 Mbps to 1 Gbps Internet access services grew the fastest, from 2010 to 2012, according to a new study by the  National Telecommunications and Information Administration (NTIA). Though growing from a low base, availability of 1-Gbps services grew nearly 300 percent between 2010 and 2012.

For Apple, "Winning" is Device Sales; For Google "Winning" is App Usage

There is some value in knowing the percentage of all mobile phones that are "smart," because that tells us something about the shape of "computing." But over time, all phones will be "smart," so the importance of "smart phones," as compared to "use of phones," ultimately will prove less relevant. 

Nor, in one sense, do all suppliers approach the "smart phone" market the same way. 
For Apple, which always has made money selling devices, revenue and profit margin matter. Samsung's concerns are similar. 

For Google, what matters is the base of users of its applications. And despite the ISP concern about "share of revenue" within the Internet ecosystem, one fact remains. As has been the case since the advent of the dial-up access market, the overwhelming share of revenue earned within the ecosystem is garnered by access providers. 






Friday, May 24, 2013

Disruptive Internet Access

Internet service providers many places in the world must operate disruptively, simply because they cannot afford to invest as much as has been the case in the tier one service provider business and also typically must operate at challenging levels as well.

That means creating services at vastly lower costs, faster than possible in the past, and sometimes designing networks that operate very differently from networks traditionally built to provide communications services, as well as operating at far lower costs.

None of that is to deride the role mobile service providers are likely to play as providers of widespread Internet access in virtually all markets. But mobile access (coverage) also typically must be augmented by high-capacity access as well.

In some cases, that might mean non-traditional antenna platforms, such as balloons, could be part of the solution. Space Data, for example, has been using such an approach to provide telemetry services to customers in the energy and transportation business.

Space Data also supplies its platforms for military applications. Google reportedly has seen promise in the approach.

Other technologies could be important as well. But some believe the balloon approach could be valuable in developing regions, functioning cheaply, but almost like low orbit satellites.

People have been looking at using balloons for communications for at least a decade.

For some, the approach will be unsettling. The balloons are unguided. They can stay online for perhaps 24 hours before self-destructing. But many disruptive technologies are like that. And the normal path of development is for a disruptive technology that doesn’t offer the full advantages of a legacy technology to progress, gradually adding more and more of the features of the legacy service.

If balloons can create a bigger market for retail Internet access, then the rest of the backbone infrastructure also will improve to meet the demand. And that better backbone infrastructure will allow additional local access platforms to reach commercial thresholds.


That's one element of disruptive Internet access: extending access at a minimum level. The other element is disruption of the high end of the access business. It appears Google will work both ends of the spectrum.

Thursday, May 23, 2013

Price, Not Availability, is 100-Mbps Problem


The real problem with very high speed (100 Mbps or higher) Internet access in the United States is the cost, not the availability, one might argue.

Some 82 percent of U.S. homes can buy Internet access of speeds of at least 100 Mbps, from a cable operator. About 18 percent of homes can buy fiber to home services.

Vectored DSL might allow many telcos to offer 50 Mbps to 100 Mbps, even if fiber is not pulled all the way to the home.

At least 235 U.S. service providers offer services at 50 Mbps or more, 136 offer 100 Mbps or more and 64 offer gigabit speeds. Most of the 100 Mbps and gigabit providers focus on the business market, given the retail prices, it is fair to say. But most cable operators happily sell 100 Mbps to consumers.

The issue really is price. Not many consumers are willing to pay $110 for a 100-Mbps connection bought with a triple play offer, or $200 on a standalone basis. In the past, such services cost as much as $300 a month.

The good news is that prices per Mbps prices per Mbps have been dropping as headline speeds have grown. Google Fiber has a cost per Mbps of about seven cents, where many offers have costs of $2 or more per Mbps of speed.



Gigabit Access Availability Grew 300% from 2010 to 2012

The availability of 100 Mbps to 1 Gbps Internet access services grew the fastest, from 2010 to 2012, according to a new study by the  National Telecommunications and Information Administration (NTIA). Though growing from a low base, availability of 1-Gbps services grew nearly 300 percent between 2010 and 2012.

Availability fo 100 Mbps services grew even more: 448 percent between 2010 and 2012. Availability of 50 Mbps services grew 160 percent between 2010 and 2012.

Services operating at 25 Mbps, arguably the speeds most consumers tend to buy, grew about 57 percent, in terms of availability.

Availability of lower-speed services has reached virtual ubiquity. Some 98 percent of U.S. residents can buy Internet access at speeds of 3 Mbps or greater and upload speeds of 768 kbps or greater.

About 91 percent of U.S. residents can buy access at 10 Mbps downstream. Some 78 percent can buy access services operating at 25 Mbps downstream.

Also, about 81 percent of U.S. residents can buy mobile broadband access at speeds of 6 Mbps or greater.

And nearly 26 percent of the population can buy fixed wireless service with download speeds at 6 Mbps.

Up to this point, cable operators have been the primary providers of high speed access services of at least 25 Mbps or greater but less than 1 Gbps. That should start to change as more telcos begin to upgrade to networks offering speeds up to 1 Gbps.

Still, at the moment, 82 percent of U.S. homes have access to speeds in excess of 100 megabits per second, while in Europe, only two percent of the population has access to these speeds, Comcast notes.



If You Want High Take Rates for High Speed Access, Price Matters

Price matters, Verizon has found. By making it easy for consumers to upgrade FiOS Quantum high speed access services to 50 Mbps for an incremental $10 or $20 a month, Verizon is seeing a "huge take rate," said Fran Shammo, Verizon Communications CFO.

"With the tools that we have with FiOS and delivering messages to our consumers on the TV or on their broadband connection, that they have the ability to just click and upgrade to 50 megabits for an extra $10, we are seeing a huge take rate on that click," Shammo said.


To be sure, Verizon often has very good reasons for wanting customers to upgrade. As it converts customers fro digital subscriber line to FiOS, it must allow consumers to keep the legacy price. So that generally means a former DSL customer gets the 25 Mbps service, for the same price as the older DSL. 


That effectively means Verizon has invested in FiOS but is unable to reap any incremental revenue for doing so, in terms of high speed access. The upgrade offer allows Verizon to generate some incremental revenue. 


In fact, Shammo suggests the upgrade policy is driving four percent growth in consumer revenues, year over year. 



Shammo also says Verizon believes use of Long Term Evolution in place of fixed access can work in rural areas, but not in urban areas. What Verizon found from its tests of LTE-plus-DirecTV was that data consumption was so high that mobile network performance suffered.

The other issue is that LTE will not deliver the 50 Mbps or 100 Mbps access Verizon believes will be needed in urban areas. 

The larger point is that price matters, when ISPs try to migrate consumers to higher-speed service plans. Take rates for 50 Mbps, 100 Mbps or 1-Gbps service plans can be significant, when retail prices are deemed to offer value. 


The point is that take rates will not be too high when prices are in the $100 a month rate. Take rates climb dramatically when the cost of a much-faster plan is only $10 or $20 a month over the existing basic plans. 



Where it comes to very high speed Internet access, price really does matter.

In August 2000, only 4.4 percent of U.S. households had a home broadband connection, while  41.5 percent of households had dial-up access. At that time, the effective price for a 1-Mbps connection might have been $234.

A decade later, dial-up subscribers declined to 2.8 percent of households in 2010, 68.2 percent of households subscribed to broadband service, with effective prices per Mbps of perhaps a couple to a few dollars.

That suggests what will happen, eventually, with take rates for 50 Mbps, 100 Mbps or 1 Gbps services. As prices per Mbps of service drop sharply, take rates will climb rapidly.

Up to this point, most consumers have not felt the need to upgrade to the fastest available speeds, in part because retail prices reflect costs per Mbps of service that are seen as “too high.”

In the U.K. market, for example, though service at 30 Mbps is available to at least 60 percent of homes,  buy rates were, in mid-2012, at about seven percent (to say nothing of demand for 100 Mbps).

But price arguably has much to do with the resistance. By June 2012 about 75 percent of U.S. households could buy a service of at least 50 Mbps, while half could buy service at 100 Mbps. Relatively few chose to do so.

For the sake of argument, assume a price per Mbps of $3. That implies a 20 Mbps connection would cost about $60 a month. Typically, the faster a connection is, the lower the per-Mbps price actually becomes. So assume a $1.50 per Mbps price for 50 Mbps services. That implies a monthly price of $75  a month.

A 100-Mbps service might sell for about $1 per Mbps, implying that a 100-Mbps service costs $100 a month.

Google Fiber, of course, deliberately disrupts those pricing metrics, offering 1,000 Mbps for $70, or seven cents per Mbps.

When price per Mbps of service drops two orders of magnitude, most people will upgrade to much faster service.





Wednesday, May 22, 2013

Top Countries for Mobile Payments Potential


Singapore, Canada, the United States, Kenya, South Korea, Japan and the United Kingdom are among the countries where mobile payments are poised to get traction, according to Business Insider.


The index measures the ability to completely replace use of credit and debit cards with mobile devices. But each of the countries ranks high in terms of readiness for different reasons.

Access is where Singapore shines. It  has 100 percent mobile coverage, while 70 percent of the nation’s residents have internet access and 68 percent own a mobile phone.

Canada, according to MasterCard, has the best cooperation among mobile networks, financial institutions, and government agencies. Plus, Canada is the global leader in Internet penetration with 82 percent of consumers online.

The United States has the most mobile-payment potential, in terms of household spending per capita, at $33,000, the highest in the world.

Kenya already has 68 percent of residents frequently using mobile payments, more than in any other nation.

Category Leader Kenya

South Korea is notable because Koreans use mobile payments to transfer money between two individuals about three times more often than they do to make point-of-sale purchases.

Japan has potential because, on average, each person in Japan has 14.98 credit or debit cards, creating a huge opportunity for displacement.



The U.A.E. is the global leader in business adoption of new technologies.

Household Consumption Expenditures

The United Kingdom makes the top of the list because of its extremely high household mobile-phone (85 percent) and Internet penetration (80 percent).

Saudi Arabia has potential because more consumers who use mobile payments are making point-of-sale purchases in Saudi Arabia than in any other country.

China potential comes from its sheer population size.

Negroponte Switch for Mobile Traffic?

To the extent that tablets and smart phones are displacing time formerly spent with PCs, there is potential for a shift in demand for access from mobile and fixed networks. That's what offloading of mobile traffic to Wi-Fi networks is all about. 

On the other hand, users also are shifting application use formerly occurring on PCs to their smart phones and tablets. So some traffic formerly on the fixed network is loaded to the mobile network. 

But the composition of the traffic is uneven. Where once television was delivered "over the air," it has over the past several decades moved to the fixed network, while "narrowband" traffic (voice and messaging) has moved to the air (mobile networks). 

Some might remember that this was known as the "Negroponte Switch." 

Something along similar lines is happening with "mobile" traffic. The bandwidth-intensive applications, especially video, is moving to the fixed network. The real-time communications traffic (chat, messaging, voice) and real-time application traffic (navigation, some transactions) is moving to the mobile network. 

"Wi-Fi-Only" Works About Half the Time: Issue is What Might be Possible in Future

Historically, the big value mobile phones represented was the untethering of "calling" from places. In fact, so great was the value that mobile calling displaced less expensive place-based calling. 

More recently, the value of a BlackBerry was the ability to use email anywhere. In the latest iteration of the untethering trend, people now expect to be able to get access to the Internet anywhere they are. 


But there is another trend happening as well, namely that people find it useful to shift Internet app consumption to stationary or fixed modes, to avoid mobile Internet access charges. In other words, mobile networks are for mobility, fixed networks are for capacity

And since perhaps 80 percent of Internet data consumption now occurs "indoors," capacity increasingly has become the value, rather than mobility, even for mobile devices. 

That is not to say nomadic access is not important, only that essentially tethered access has become more important. Just how important remains to be seen. 

Generally, one might say that "synchronous" or "real time" communications and apps benefit from "anywhere, anytime" access. Asynchronous apps and communications (email, voice mail, blog and social posts) can tolerate some periods of disconnection, and are better suited to place-based access. 

So the issue is the degree to which growing use of asynchronous apps means access can be useful if it is "not always connected," as is the case for any user relying on devices with Wi-Fi access only, and not full mobile network access. 

At least so far, voice and messaging devices (phones) have benefited from "anywhere, anytime" access. PCs and now tablets often can provide high value even when connectable only sometimes. 

The big zone of uncertainty is whether smart phone Internet access demand will change to any great extent. The possible change is voice and messaging on the mobile network, with Internet access mostly or even exclusively based on Wi-Fi access. 


Smart phone owners know they can use Wi-Fi access inside and outside the home, and there is a financial benefit to doing so.


Mobile service providers also realize they can avoid capital investment by encouraging their users to switch their devices to Wi-Fi whenever possible, as well.

Voice-based or communication-based applications generally are not the best candidates for “Wi-Fi-only” networks. But tablets, PCs and Chromebooks are a different matter. And smart phone Internet access already is a case where mixed access is the norm. That already is allowing creation of business models based primarily on fixed access (Wi-Fi) with a mobile overlay.
34 Percent of Global Tablets will be Cellular Connected by 2017

BT, for example, appears to be thinking along those lines.

Having won 4G spectrum (2x15 MHz of FDD and 20 MHz of TDD 2.6GHz spectrum), BT suggested it would not build a retail mobile network, but use 4G to augment BT's fixed networks.

Now BT says it will launch its own retail network.

The thinking is that BT will source wholesale mobile connectivity from one of the U.K. mobile service providers to provide full mobile access, while using its own spectrum largely for fixed or location access.

That is analogous in many ways to the ways mobile service providers already blend full mobile access plus Wi-Fi access. The potentially big challenge is whether it might eventually be possible to create 
access services that have high value even if there is no mobile access, as once was thought feasible around the turn of the century. 



Those of you who travel outside your home country already do this: you turn off your mobile Internet access and rely only on  
Wi-Fi  
when out of country. 







"Tokens" are the New "FLOPS," "MIPS" or "Gbps"

Modern computing has some virtually-universal reference metrics. For Gemini 1.5 and other large language models, tokens are a basic measure...