Headlines in Tech 23-29 Mar 2022

Tech News Pick of the Week

Digital Markets Act, EU’s weapon against the BigTechs

The Digital Market Act is now out. It will concern significant platformers that control “core platform services” which meet certain thresholds. It’s really an act targeted at GAFAM (Google, Amazon, Facebook, Apple and Microsoft) – though some EU players may also be in the frame [it is suggested that the thresholds were lowered to ensure that those that are subject are not all those from the US]. Those qualifying are called Gatekeepers. They are not happy. Serious negotiations ongoing. 

Minimum you need to know

  • The agreement reached between Council and the European Parliament is provisional and subject to approval by the same. Expected to come into force October [though the rules will apply to the companies 6 months after that]. 
  • Tougher than expected
  • Core platform services include marketplaces and app stores, search engines, social networking, cloud services, advertising services, voice assistants and web browsers
  • Certain obligation on emerging gatekeepers
  • Huge fines, up to 10% of worldwide turnover [note: not profits] and up to 20% for repeat offenders [so what Apple’s doing right now, with the finding (currently on appeal) that it has breached Dutch competition law in the online dating app market (Apple prohibits in-app purchases via third party payment system across the board but the finding only bites on online dating market only), is to pay the penalty amounting to €5 million per week rather than make changes to its system – will become untenable]. 
  • There is also available structural remedies [eg. unwinding acquisitions, splitting companies up –it is said that companies like Amazon can undercut all its e-commerce competitors by operating without much profit because it can fuel its operations from the profits from elsewhere, such as the cloud service AWS –and therefore, query has been raised whether Amazon should be split up]

Gatekeepers have the positive obligation to do the following:

  • Users must be able to unsubscribe easily as subscribing [no more “dark patterns” such as “if you leave, you will no longer get X Y and Z, are you sure?]
  • Refrain from pre-installing important software (eg. Chrome) by default upon installation of an operating system. [Google pays Apple to make it iOS search engine default – this practice is likely to be prohibited under the DMA – there is currently US class action for breach of competition law – the deal between the two is said to incentivise Apple not to make its own search engines]
  • Ensure interoperability of instant messaging services’ basic functionalities [so you need to be able to send group messages, such as Apple’s imessage to those that have an Android device – or you will be able to send an WhatsApp message to a Facebook Messenger user – but serious issue is technically, how is this achieved? What if the encryption methods are different? Or proprietary?] In short, the DMA tries to make the ecosystem work more like the Internet where I can send an email from my Hotmail account to someone with a Gmail account – this is all doable because the protocols for internet messages are standardised. 
  • Allow App developers to have access to supplementary functionalities of smartphones (eg. NFC chip)
  • Sellers to get access to marketing performance data 
  • Inform the EU Commission on mergers and acquisitions. 

Gatekeepers can no longer do the following:

  • Self-preferencing [Amazon can’t put their own products at the top when the user searches for an item]
  • Re-use private data collected for one service for another service [So Google can’t use the geolocation collected to advertise eg. swimwear on the side bar of Google search, where Google knows the user is a keen swimmer, or about to go on a holiday to a tropical destination]
  • Impose unfair conditions for business users
  • Require app developers to use certain services [can’t force users to pay by Apple Pay, for purchases made on a third party app].  

Did you know?

There is a new acronym for the US semiconductor giants, MANGO, that stands for Marvell, AMD, Nvidia, Global Foundaries, On Semi. Bonus points to you if you wondered where Intel is. I am not sure. 

Artificial Intelligence

Google Parent Alphabet’s quantum software unit Sandbox AQ spins out

…Sandbox AQ is a software company providing solutions at the intersection of Quantum Computing and Artificial Intelligence. The use cases for Sandbox’s software are in the following areas that require high performance computing power:

  • Cybersecurity
  • Healthcare (drug development)
  • Energy and the environment
  • Telecom
  • Financial services
  • Government

BigTech

Google allows third party apps to use its own in-app payment system

…Spotify is trialling its own payment system on Google’s app store. Surely there’s added pressure on Apple. Google and Apple enjoy enormous benefits from the 30% commission it gets from users making in-app purchases including subscriptions.  The question is how much commission Spotify will have to pay when using a third party payment system on the Google platform. 

Following the fall out from the dispute between Match.com and Apple in the Netherlands, Apple suggests that such commission should then be reduced from 30% to 27% to account for the fact that Apple is no longer providing the payments services (ie: removing the sums related to payment processing and related activities – with developers responsible for collection and remittance of taxes & VAT – it may well be cheaper for the developers to accept the 30% commission). 

Google has reduced fees for smaller companies to 15%. Microsoft store charges 12%.

Yesterday’s enemy is Today’s friend: New York yellow cab drivers agree to be listed on Uber

…I do hope Uber will make friends with the London Taxis soon. It has been impossible to get cabs in the city centre these days. Uber will get a commission and be able to meet the demand and the legacy taxi drivers will receive intel on where supply is likely to arise and serve Uber’s users. It’s a win-win for all.

FTC probes Microsoft’s proposed acquisition of Activision Blizzard

…There is no surprise here really. Microsoft proposes to buy the third largest gaming company for about $70billion. To increase the chance of the acquisition being accepted, Microsoft has made some promises to try and pre-empt any fierce regulatory interventions:

  • Open App Market Principles laid out for its gaming store [which overlaps a lot with the Digital Markets Act]
  • Developers can use third party payments systems
  • Own products won’t be given preferential treatment
  • Refrain from using private information from third party apps to sell its own software
  • Coming into agreements with Sony and Nintendo so that they may offer popular Activision Blizzard titles on their platforms.  

Microsoft’s cloud service (Azure) hit with antitrust complaint in the EU

…Complainant is OVHcloud, a French cloud service provider. There are two unidentified complainants alongside. The allegation is as follows:

  • Microsoft’s licensing arrangement makes it more expensive for user to switch between cloud providers
  • Microsoft software does not perform well on non Azure cloud platform

Microsoft is subject to another complaint, by NextCloud (a European company) which complains of Microsoft’s bundling of Onedrive and other Microsoft services. 

Cloud market

Microsoft Azure, Amazon’s AWS and Google’s Google Cloud account for nearly 70% of the European cloud market. 

Deutsche Telecom is Europe’s top cloud provider, which accounts for 2% market share, and OVHcloud accounts for 1%. 

Data

US – EU Agree Transatlantic Data Privacy Framework – in principle

…In 2020, CJEU ruled that US surveillance practices in the US meant that EU citizens’ data could not be adequately protected, such that businesses have thus far been prohibited from transferring personal data of EU citizens to certain non-EU countries with inadequate protection under GDPR. The new agreement seeks to allow for data to flow between the regions, as a result of certain changes to US law and practice, ensuring that government access to EU citizens’ data is proportionate and restricted to instances to only where necessary. 

EV

Jaguar Land Rover in talks with battery supplier Envision to set up battery gigafactory in the UK

…Envision makes car batteries and supplies to Nissan. The take away point is that the OEMs based in the UK are setting these factories for parts, in the UK. This could be to reduce supply problems, and the whole concept of striving for national self-sufficiency. It is said that the UK is preferred although Hungary and Spain are also in the running. In other words, it could be a sign of de-globalisation.

Fintech

Apple buys fintech start up Credit Kudos (UK)

…Credit Kudos, lending platform using alternative credit scores (bit like Tesla’s motor insurance offering then – it uses the driver performance collected by the Tesla car to decide the premium). It suggests Apple will be getting into banking services, beyond Apple Pay (eg. general loans, or even mortgages, student loans?). With the information gleaned from Apple Pay, Apple can decide who to loan money, and at what rate [or does that infringe the provisions of the Digital Markets Act I wonder – you aren’t allowed to collect data for one service and use for another]. Rumours also suggest that Apple will get into BNPL (Buy Now Pay Later). Apple pay already does Apple cards which enables users to BNPL for Apple devices (ie: buy iPhones with 0% interest).

Litigation

Google accused of training employees to deliberately make documents privileged when they need not be

…The US DOJ says that Google has been instructing employees to label documents as being privileged and seeking in-house counsel’s advice on a high level basis even where no advice is needed.  Email chains show that in-house lawyers do not respond to the questions asked of them, leading to the allegation that in-house legal involvement is engineered in an effort to manufacture false privilege claims. Google is said to have provided this arrangement under the moniker “Communicate with Care”, stipulating that the employees ought to add an “artificial indicia of privilege” to documents, including those that concern revenue sharing and distribution agreements. 

Metaverse / AR / VR

Snapchat owner Snap buys NextMind

…Nextmind is a “neurotech” company concerning brain/computer interface, based in Paris. Its ultimate aim is to enable the control of VR and AR headsets, hands-free. It monitors neural activity in your brain allowing you to push a virtual button. Snap is developing the VR and AR headset Spectacles, which is subject of a trade mark litigation in the US (vs TM issuing office, the USPTO). 

It is evocative of Elon Musk’s Neuralink which is an implant designed for your brain that allows you to control iOS, keyboard and mouse just by thinking about it. 

Semiconductors

Nvidia throws down the gauntlet as it announces a new Arm based chip

…It’s a challenge against Intel and AMD as the chip is designed to have a level of performance which can be used  for the CPU (central processing unit, ie: the brains of a device) to power the data centres. Nvidia’s main objective behind the ARM acquisition has been believed to be to enter the data centre CPU market. They appear to have done that without having to acquire ARM.

Nvidia are investing heavily in making an “operating system for AI”, and with it software so that its GPU (Graphics Processing Unit, the bit that does very complicated processing, such as graphics rendering and AI data processing involving huge volumes of data) can be used more widely. It has developed the Cuda software which enables developers to use new forms of advanced computing based on its GPUs, in particular in the area of autonomous vehicles and metaverse. 

Trade Secrets

Trade secret misappropriation through acquisition of company

…Most trade secret misappropriation cases are about employees stealing trade secrets of an employer and then transferring that to a rival company. Sometimes you also get companies working on some joint venture or collaboration and one of the collaborators use knowledge gained from the joint activity to use it for a different purpose. But this is a case where there is an allegation of trade secret misappropriation that have arisen as a result of company acquisition.
Aristotle, a data mining company, had bought the services of Acuant Inc, a fraud prevention company. Acuant sought confidential information (to do with layered software security, among other things) from Aristotle, but unbeknown to Aristotle, Acuant was in discussions for an acquisition by Aristotle’s rival, GB Group. Aristotle says it would not have transferred the confidential information were it made aware that Acuant was being bought by its rival.

Delving Deeper

Drug discovery and Artificial Intelligence

…Drug discovery is horrifically expensive. I remember learning at some point – I cannot remember the source – that the average cost of drug discovery is $2billion, when you account for the fact that the great majority of candidate drugs fail to make it (either they are not effective or not toxic). I am sure there are variations on the figures but no doubt that the sum is vast, however you cost it out.
AI could hold the key to cutting down that cost. But how is it done?
First phase
AI can be used to narrow down the number of drug candidates, based on inputs provided by the scientists to identify a candidate which is likely to succeed in clinical trials.
Second phase
Robots are then used to carry out the routine experimental process. They carry out the routine experiments and record the process and the result.
Third phase
AI can be used to identify the right patients for the drug.
Examples in the past
Benevolent AI – trawled through 50 million medical journals searching for approved drugs that could be repurposed for other diseases (Jan 2020).
Dai Nippon Sumitomo Pharma – Puts drug DSP-1181, a treatment for obsessive-compulsive disorder to test after having identified it using AI – with Exscientia (Feb 2020)
Eli Lilly – AI was used to narrow down the candidates to barcitinib – a drug for rheumatoid arthritis, taking only 4 days for the narrowing down process, and then put to clinical trial. Barcitinib was already known, but again the drug was identified as being helpful for repurposing (Jul 2020).

Microsoft goes one step further on Confidential Computing

…So what is Confidential Computing? 

Everyone uses the power of cloud at the moment. Businesses will collate huge amounts of data, upload into the cloud for storage or further processing. Data that is uploaded can be massively confidential – my personal email account contains sensitive information about my health and finances for example. The problem is even more acute for hospitals and banks who has control of such data pertaining to a large number of people.

In order to ensure privacy and security of such data, confidential computing has become essential. It:

  • Isolates user data in the computer memory, thereby keeping it unexposed to operating systems, applications and different users of a cloud server
  • That part of the memory (Trusted Execution Environment, or TEE) is only accessible by certain computer programs which can read and write data. 
  • Data in the TEE is encrypted always – during transit, rest, when in use.

Intel and AMD are implementing TEE in their CPU, Microsoft offers this in their cloud computing offering, Azure. Microsoft is now teaming up with Nvidia to apply the concept to GPUs, which are used in more complicated data processing (typically, graphic processing and AI – it can carry out large number of complicated calculations in parallel), and typically work with CPUs (which isn’t as fast as GPUs, but can carry out more instructions, or commands) to carry out whatever high order task it is set to do. Nvidia will be developing GPUs that can work with CPU TEE, ensuring encryption end to end, even when GPUs are involved.