$36 Billion in AI Spending in 2019

Firms and agencies in the United States will represent nearly 66 percent of all spending on artificial intelligence systems in 2019, or about $23 billion, led by the retail and banking industries, IDC analysts predict.

Western Europe will be the second largest region in 2018, led by banking, retail, and discrete manufacturing, IDC says.

The strongest spending growth over five years will be in Japan (58.9 percent compound annual growth rate) and Asia/Pacific (excluding Japan and China) (51.4 percent CAGR). China will also experience strong spending growth throughout the forecast (49.6 percent CAGR), according to IDC.

source: IDC

Worldwide spending on artificial intelligence (AI) systems is forecast to reach $35.8 billion in 2019, an increase of 44 percent over the amount spent in 2018, IDC predicts.

IDC also expects spending on AI systems will more than double to $79.2 billion in 2022.

Global spending on AI systems will be led by the retail industry where companies will invest $5.9 billion this year on solutions such as automated customer service agents and expert shopping advisors and product recommendations.

Banking will be the second largest industry with $5.6 billion going toward AI-enabled solutions including automated threat intelligence & prevention systems and fraud analysis & investigation systems.

The industries that will experience the fastest growth in AI systems spending over the 2018 to 2022 period are federal/central government (44.3 percent CAGR), personal and consumer services (43.3 percent CAGR), and education (42.9 percent CAGR), IDC believes.

The AI use cases that will see the most investment in 2019 are automated customer service agents ($4.5 billion worldwide), sales process recommendation and automation ($2.7 billion), and automated threat intelligence and prevention systems ($2.7 billion).

Five other use cases will see spending levels greater than $2 billion in 2019: automated preventative maintenance, diagnosis and treatment systems, fraud analysis and investigation, intelligent process automation, and program advisors and recommendation systems.

Software will be the largest area of AI systems spending in 2019 with nearly $13.5 billion going toward AI applications and AI software platforms. AI applications will be the fastest growing category of AI spending with a five-year CAGR of 47.3 percent.

Hardware spending, dominated by servers, will be $12.7 billion this year as companies continue to build out the infrastructure necessary to support AI systems.

By the end of the forecast, AI-related services spending will nearly equal hardware spending.

AT&T Shift to Streaming Linear TV Has Numerous Advantages

As some had speculated or feared, AT&T’s purchase of DirecTV is leading to a major change in video entertainment delivery, away from linear service based on the fixed network, away from satellite delivery, and towards streaming.

There are all kinds of implications. Not the least of the advantages are operating cost reductions. “The biggest cost we have it that is so to speak, the truck role and getting that installation out,” said John Stephens, AT&T CFO. So AT&T has been testing a self-install decoder “called Osprey,”  which is a “self-installed, full linear product.”

So “the only truck roll is the UPS truck,” he quipped. “It dramatically reduces our install cost; dramatically reduces our subscriber acquisition costs.”

So the standard linear video product shifts from U-verse to DirecTV to DirecTV Now, using a self-install decoder and “bring your own broadband.”

The Osprey is said to be an Android-TV-powered streaming player that will offer the same linear service contracts and channels but without the use of a satellite dish.

That deployment model keeps DirecTV’s national footprint, but shifts the platform to streaming. In principle, that also gives AT&T more options about how to upgrade internet access bandwidth inside its fixed network footprint.

The shift to streaming eliminates the need for a dedicated linear video network. By unbundling access and app, AT&T also gains the ability to use any access platform (its own, or others) to support linear streaming video. Where U-verse video segregates linear video from internet access, the streaming platform shares the access pipe.

Oddly enough, AT&T might have to provision less bandwidth for linear streaming than for U-verse video, as U-verse video used separate logical networks over the same access cable.

As was the case for cable TV operators migrating from analog video to digital video, one important advantage was that additional bandwidth was freed up for internet access purposes.

What are Comcast's Mobile Objectives?

Not everyone might agree that Comcast has no intention of ever competing with AT&T and Verizon, as Comcast now claims.

Cable has in the past used a “crawl, walk, run” approach to new services, ranging from voice to business services. So it makes sense to disavow larger ambitions in the early going.

Nor might it ultimately matter that Comcast now is losing money on mobility services, as all observers seem to agree. The Xfinity mobile service is not profitable at the moment.

Of course, initial losses are not uncommon for any service provider launching new services of any type in the telecom business.

Some of that arguably is due to scale. Profits are hard to come by whenever a mobile service provider has a small subscriber base, and Comcast says its mobile service is not yet profitable.

And though Comcast might be growing its gross additions at up to a 20-percent rate per quarter (Comcast has not reported net gains), it had something more than 1.2 million accounts at the end of 2018.

Though some estimate cable operators might by perhaps 2020 be getting half of all industry net account additions, that is a ways off.  

Startup costs likely also are an issue (marketing, phone installment plans, wholesale capacity purchases and operations).

But scale might not be the chief initial issue. Actual acquisition costs might be running between $1600 and $1800 per new account, even if Comcast has reported acquisition costs of about $1260 per new account, according to one analysis by BTIG.

Consider that Tracfone, which also operates an a mobile virtual network operator, with perhaps 21.7 million subscribers, generates earnings (cash flow) of about 10 percent on that base of customers.  

Eventually, owner’s economics are going to look more appealing. And that means a move away from operating as an MVNO, and towards facilities in some way. So though Comcast and other cable operators might be adding to MVNO numbers right now, it remains unclear whether they will do so in the future.

Some researchers believe the U.S. mobile virtual network operator business will grow in coming years, while others suspect it will decline. It might be argued that the biggest influences will be cable operator entry into mobility and the level of competition within the U.S. industry, which seems to encourage former MVNO customers who are price conscious to switch to one of the leading four national service providers.

Global MVNOs seem to be losing accounts, overall, though growth continues in some markets.

Most can agree that MVNO accounts peaked in the U.S. market about 2012, and in 2018 represented less than five percent of all U.S. mobile accounts. One supplier, Tracfone has perhaps 65 percent market share of MVNO accounts.

source: One Development  

source: Grand View Research

source: Strategy Analytics

SD-WAN Market Grew 28% in 4Q 2018

Software-defined networking (SD-WAN) software revenue, including appliance and control and management software, rose 26 percent quarter over quarter to reach $359 million in the fourth quarter of 2018, says Josh Bancroft, IHS Markit senior research analyst.

Those are the building blocks for service provider SD-WAN services, which Vertical Systems Group has estimated at about $282 million in managed service revenues in 2018.

Providers actively selling managed SD-WAN services in the U.S. include the following companies (in alphabetical order): Aryaka, AT&T, CenturyLink, Cogent, Comcast, Fusion Connect, GTT, Hughes, Masergy, MetTel, Sprint, Verizon, Windstream and Zayo, Vertical Systems says.

source: Vertical Systems

VMware led the SD-WAN software market revenue share with 20 percent, followed by Cisco at 14 percent and Aryaka at 12 percent, IHS Markit says.

For the foreseeable future, both direct and channel sales will continue to drive SD-WAN market growth.

Respondents to IHS Markit surveys have shown a preference for self-managed SD-WANs using on-site hardware and software. But managed services are gaining share.

source: IHS Markit

The Trouble with Antitrust: Past, Present, Future

Antitrust issues are at the forefront of the proposed merger of T-Mobile US with Sprint, but also with concerns about the market power of firms such as Facebook and Google.

The point of antitrust legislation is consumer protection. So the empirical issue is whether antitrust action actually works, in practice. Do such actions (breakups, for example, as some suggest for Facebook and Google) protect consumers and provide benefits such as lower prices?

The answers might not always be clear. Firm actions are not always unambiguously pro-competitive or anti-competitive.

“When prices decline sufficiently so that no firm in an industry is earning economic profits and some firms exit, this outcome may reflect a highly competitive market,” note economists Robert Crandall and Clifford Winston of the Brookings Institution.

In other cases, a large competitor might be engaging in predatory pricing to drive out its rivals.

If one contestant adds lots of capacity, is that an example of an implied threat to cut prices drastically, or a way to add more competition?

“Almost any action by a firm short of outright price fixing can turn out to have pro-competitive or anti-competitive consequences,” the economists notes.

After studying the impact of antitrust actions, they say they can “find little empirical evidence that past interventions have provided much direct benefit to consumers or significantly deterred anti-competitive behavior.”

So “authorities would be well advised to prosecute only the most egregious anticompetitive violations,” they note, as such actions are inconclusive. Looking at the big textbook examples of major antitrust action, they find

* the breakup of Standard Oil had little effect on either consumers or on profits
* The American Tobacco case did little to spur meaningful competition
* the decree did not reduce Alcoa’s dominance
* price of a movie ticket rose in the two decades following the Paramount decision and there was little market entry by new competitors
* After antitrust action, USM’s revenue gains were more than twice the sum of its four major competitors’ gains
* After the AT&T breakup, long distance prices did fall, but that is attributable to “equal access” rules, not the breakup
* Monopoly cases against Safeway and A&P had little impact on market structure
* The charge of American Airlines predatory pricing at hubs is inconclusive

One generic problem is that antitrust cases take so long to conclude that industry structures often already have changed by the time a decision is rendered. That was true in the Standard Oil, Alcoa, IBM and Microsoft cases, they say. In other cases, including American Tobacco, Paramount and United Shoe Machinery cases, negligible consumer price or industry structure changes happened.

Also, mergers may harm or benefit consumers. It depends. Mergers that enable firms to acquire market power may only raise consumer prices, while mergers that enable firms to realize operational and managerial efficiencies can reduce costs and thereby lower prices, the economists notes.

After reviewing about a hundred merger reviews and cases, the authors say “regulators are not sorting out good mergers from bad ones with much accuracy” and that “antitrust authorities overreach and attempt to block productive mergers.”

“We can only conclude that efforts by antitrust authorities to block particular mergers or affect a merger’s outcome by allowing it only if certain conditions are met under a consent decree have not been found to increase consumer welfare in any systematic way, and in some instances the intervention may even have reduced consumer welfare,” they say.

“We have not found any evidence that antitrust enforcement has deterred firms from engaging in actions that would have seriously harmed consumers,” they add.

In the end, “any deterrent effect of the antitrust laws may be relatively small compared with the well demonstrated ability of competitive markets to deter anti-competitive monopolies, collusion and mergers,” they conclude.

The point is that antitrust actions often fail to achieve the objectives of greater competition and consumer price benefits.

“The apparent ineffectiveness of antitrust policy stems from several causes:
1) the excessive duration of monopolization cases, which portends that the particular issue being addressed will evolve into something different—often of less importance—by the time it is resolved;
2) the difficulties in formulating effective remedies for monopolization and effective consent decrees for proposed mergers;
3) the difficulties in sorting out which mergers or instances of potentially anticompetitive behavior threaten consumer welfare;
4) the substantial and growing challenges of formulating and implementing effective antitrust policies in a new economy characterized by dynamic competition, rapid technological change and important intellectual property (Carlton and Gertner, 2002);
5) political forces that influence which antitrust cases are initiated, settled or dropped (Weingast and Moran, 1983; Coate, Higgins and McChesney, 1995), including situations where firms try to exploit the antitrust process to gain a competitive advantage over their rivals (Baumol and Ordover, 1985);
6) the power of the market as an effective force for spurring competition and curbing anticompetitive abuses, which leaves antitrust policy with relatively little to do.”

There are some who say the emphasis on consumer welfare has to change. Bigness itself now seems to be the rationale for action. In what some claim is a return to earlier principles, the argument is that a focus on consumer welfare, especially prices now is inadequate.

Instead, antitrust should focus on potential harm to competitors, reduced innovation, jobs, reduced market entry, decreases in product quality, and privacy.

Since traditional antitrust has been relatively ineffective, one wonders how it might be any more effective when a number of harder-to-measure outcomes are substituted. We have done poorly enough when quantitative measures were used; we are likely to do far worse when qualitative measures are the substitute.

T-Mobile US Promises Mobile Substitution for Fixed Internet Access

Perhaps the most-startling new argument T-Mobile US is making in support of its merger with Sprint is the ability to use the 5G network to provide the equivalent of fixed network internet access without using fixed wireless, relying solely on the mobile network to reach about half of all U.S. homes.

The “broadband in a box” solution that New T-Mobile does not appear to require use of an exterior antenna of any sort, the company says in a new filing with the Federal Communications Commission.

“New T-Mobile will simply ship a box to the customer’s home for the customer to self install,” says T-Mobile US.

That claim also suggests the network supporting the offer will not be “fixed wireless” but the mobile network itself, a use case that has become a reality in the 4G era for lighter users. But T-Mobile US seems to suggest the 5G mobile network will be a platform for full competition with fixed internet access services, generally.

source: T-Mobile US

By 2024, T-Mobile expects to cover about half of U.S. homes with the in-home Internet service, relying on the mobile network, not fixed wireless, providing average download speed in excess of 100 Mbps or higher (with a minimum speed of 25/3 Mbps), the company says.

“New T-Mobile will use this low cost structure to aggressively capture share by pricing its service at (redacted) per month below what in-home broadband providers typically charge today,” making the story “lower prices for fixed network internet access.”

That would be the latest example of mobile substitution that began with voice services.

Is T-Mobile US Sprint Merger in Trouble?

I never claim to understand all the politics of communications law and regulation, but it does seem to me that the latest filing by T-Mobile US, and the pause of the merger review by the Federal Communications Commission, does suggest policymakers are not yet convinced the Sprint merger with T-Mobile US is really in the public interest. Indeed, merger opposition is not a new story. The AT&T effort to buy T-Mobile US was rebuffed, for example.

Initially, the claim was that the merged company would be a stronger competitor to AT&T and Verizon, which is true as far as it goes. A much-bigger company with much-greater scale--in a scale business--should do better.

Ignore for the moment the fact that no equity analyst I ever have heard from thinks “lower prices” are an outcome. Instead, they believe higher prices are the outcome. That is not to say market concentration and prices are related is a simple and obvious way.  

Still, most might agree the merger is better for Sprint and T-Mobile US. It is not so clear it is good for consumers , at least in terms of the amount of price competition in the U.S. mobile market. That noted, it remains unclear how much competition the U.S. mobile market can sustain, long term.

But some of us would argue the answer is yet unknown, as Comcast, Charter and possibly other large entities with scale could be big forces. The U.S. market arguably is big enough, with enough ways to create new business models blending app, services and access, that the “best” answer might not be “two or three” pure-play mobile service providers.

Someday, we might find that mobile access is something that is part of the core value proposition for bundles of vale that bridge apps, services, devices and network services. We just do not know where the larger business is headed, except to note that “access alone” is getting to be a dangerously-thin way to earn revenue on a sustainable basis.

I’ve argued in the past that among the most-sustainable outcomes is for Comcast and Charter to combine assets with T-Mobile and Sprint, for example, pairing one mobile firm with one cable leader. Other combinations might be proposed, but none seem to have the clean “we provide access services” rationale as well as cable-plus-mobile.

That is the analogy to AT&T and Verizon’s operation as entities with mobility and fixed assets.

The filing leads with the argument that “new” T-Mobile will challenge the cable operator “monopoly” of fixed network internet access. “New T-Mobile will not only raise the performance bar and enhance competition for mobile wireless services, but also enter into and disrupt in-home broadband,” the document says. Some of us believe that is a reasonable or plausible claim.

But it strays from the basic argument that new T-Mobile actually is good for mobile consumers. And that suggests T-Mobile believes it has not won that argument. Some of us have argued the deal would face a high bar because the U.S. mobile market already is concentrated.

In the mobile business, as well as the fixed business, competition is going to occur in different ways, in the future. So both T-Mobile US and Sprint are long-term asset sellers, if this merger is not approved. Some might argue that would be better for both innovation and competition, long term.