Context: The UK Competition & Markets Authority (CMA) has repeatedly mentioned the relevance of cloud computing resources to AI services, but its cloud services market investigation is actually focused on questions unrelated to AI (May 30, 2024 ai fray article).
What’s new: Yesterday (June 6, 2024), the CMA published several documents related to that market investigation (an updated issues statement (PDF) as well as working papers on technical barriers (PDF), software licensing (PDF) and a potential remedy package (PDF)). The part about software licensing is surprising, given that the CMA’s own qualitative market survey did not identify any hard evidence of competition issues in that regard. What is even more surprising is the CMA’s reliance, in that context and presumably others, on market leader Amazon Web Services and another hyperscaler, Google, who seek to harm Microsoft, one of their competitors.
Direct impact: In the coming weeks, the CMA accepts input from stakeholders. It will draft a decision and hold hearings later in the year. There is a possibility of the CMA still reaching the conclusion that there is no evidentiary basis for taking issues with Microsoft’s software licensing terms in connection with cloud services.
Wider ramifications: The CMA has recently appeared to be more interested in alleging competition issues in the AI and cloud contexts than in correctly applying competition law and economic logic to the facts (May 21. 2024 ai fray article). Worse still, the CMA appears biased against Microsoft to the extent that the CMA will rather shield incumbent market leaders from competition than accept the reality that in many markets, Microsoft is a challenger. Last year, the CMA made an outlier decision to block Microsoft’s acquisition of game maker Activision Blizzard in a move that would have favored no one but console market leader Sony; it has launched “merger” investigations of non-merger partnerships with AI companies; and it has now also issued a working paper on software licensing in the cloud that instead of taking a rational perspective on the evidence just adopts the arguments of the market leader (Amazon) and another major player, Google.
Last year, there was a lot of debate in British politics and in (global) media over whether the UK was open or closed for business. The most striking CMA decision that gave rise to that was the initial decision to block Microsoft’s Activision purchase. But it was far from the only one. Some companies already complained about the CMA’s heavyhandedness and arbitrariness before the Activision ruling issued, and others did shortly thereafter.
The British government made it very clear that the CMA should focus on cost-of-living issues (hint: Amazon’s or Google’s profit margins raise no such concern). That does not mean that the CMA should not enforce competition rules vigorously with respect to digital markets. The passage of the UK’s Digital Markets, Competition and Consumers (DMCC) bill into law gives the agency more powers than before. But there is no reason to assume that British politicians want the CMA to abuse that power.
On May 30, 2024, MLex’s Nicholas Hirst published an article that is no longer paywalled: UK decision not to probe Microsoft-Mistral leaves more questions than answers
In that article, Mr. Hirst quotes legal experts and shares his own observations on the CMA spelling out an arbitrary standard of review for AI and other partnerships. All sorts of commercial arrangements that were far beyond the reach of strict merger laws in the past could now be reviewed and blocked like mergers.
The CMA is, once again, on the wrong track, as if it had learned nothing whatsoever from the Activision mess. The DMCC law gives the CMA more powers, but the UK economy is still just about 3% of the global economy, and faces fierce competition from other world regions near (EU) and far.
So what’s the matter with cloud software licensing?
Obviously, there are Microsoft software products such as Windows Server, Windows for the desktop, Office and SQL Server (a database management system) that many companies use. Also, many software developers build Windows applications with Microsoft’s developer tools.
According to the CMA’s qualitative market survey, some companies consider it a logical choice to use Microsoft’s Azure cloud if they already have a large-scale business relationship with Microsoft.
What’s even more logical is that other cloud providers can’t just grant their customers licenses to Microsoft products without paying Microsoft, just like Microsoft can’t issue (at least not without paying) Amazon or Google Play (Android) gift coupons to its customers, or let its customers advertise (without paying) on Google’s search engine or on YouTube.
But Amazon (directly and through CISPE, an organization it controls) and Google have somehow managed to persuade the CMA that it should be concerned about discounts that Microsoft grants to its customers, though others grant discounts for other reasons. Amazon is still the undisputed market leader (unless one departs from long-standing market definition principles). There is no sign of Amazon being unable to compete with Microsoft, nor is there any reason to assume that Google’s fast-growing cloud business can’t compete.
The software licensing paper released by the CMA yesterday actually concedes in its paragraph 5.31 that a qualitative market survey failed to produce evidence that Amazon’s and Google’s self-serving complaints have substance:
The Jigsaw report found that individual factors did not necessarily affect participants’ attitudes towards multi-cloud or switching significantly on their own, but rather they add up and shape multi-cloud or switching behaviour in a cumulative way. The report notes that this makes the role of software licensing in the decision to go with Azure difficult to unpick, though adding that participants did not single out licensing as a key factor on its own influencing their decision. The report did highlight that pre-existing use of Microsoft was often closely related to participants’ original take up of Azure, with some participants identifying themselves as Microsoft shops.
Ignoring those facts, the CMA is now contemplating the following measures (quoting from the working paper on remedies):
(a) non-discriminatory pricing for Microsoft software products, regardless of which cloud infrastructure they are hosted on;
(b) allowing customers to freely transfer previously purchased Microsoft software products to the cloud infrastructure of their choice without incurring additional costs;
(c) increasing price transparency in relation to the use of Microsoft software products on Azure and third party cloud infrastructure;
(d) requiring parity of Microsoft software products and product functionality for use on Azure and third party cloud infrastructure.
There appears to be no justification for any of that, and the second item is actually irreconcilable with intellectual property law. Software is licensed for specific purposes. A regulator can’t just step in and give end users rights that they never paid for.
So who “planted” that idea?
Toward the end of the working paper on cloud software licensing, they discuss remedies. And almost every paragraphs begins like this:
“AWS and Google have called for remedies…”
“AWS submitted that …”
“Google has submitted that…”
“Google told us that….”
“CISPE [in other words, Amazon] has suggested that…”
“Google and CISPE submitted that…”
“Submissions from AWS, CISPE and CFSL [that’s also, practically, Amazon] advocated…”
The CMA appears more interested in protecting competitors than in safeguarding competition. It allows itself to be used as a tool by certain companies.