AI, Blockchain and the Cloud: Banks in the New Technology Era

sagawa
Print Friendly, PDF & Email

SEE LAST PAGE OF THIS REPORT Paul Sagawa / Tejas Raut Dessai

FOR IMPORTANT DISCLOSURES 203.901.1633/.901.1634

psagawa@ / trdessai@ssrllc.com

twitter.jpg @PaulSagawaSSR

May 24, 2018

AI, Blockchain and the Cloud: Banks in the New Technology Era

AI can radically improve risk management – much better outcomes at much lower cost. Blockchains can dramatically reduce the friction (i.e. cost and time) for financial transactions. Cloud platforms can offer consumer financial services with reach and convenience almost impossible for traditional banks to match. These technologies will remake the central processes of banking, creating significant opportunities for tech players to partner with, or in some cases, disintermediate incumbents as new solutions take hold. There are barriers – strict regulations will stymie many would-be tech-forward market entrants and slow the adoption of new processes by incumbents, cooperation around new interbank networks will be difficult to marshal, and customers may be reluctant to change the way that they work with their banks. Still, the benefits will be extraordinary and banks lack the wherewithal (e.g tech talent, datacenter infrastructure, etc.) to design, implement and operate the new solutions without substantial support from technology partners. This creates opportunities for tech players in three areas: Risk Management Solutions – IT consultants (e.g. IBM, ACN, etc.) will help automate processes to assess and mitigate credit, operational and market risks using AI. These systems will make use of tools, APIs and hosting provided by cloud platforms (e.g. MSFT, IBM, AMZN, and GOOGL). Transaction Networks – Banks will also rely on IT consultants and cloud platform partners to develop, implement and operate blockchains that will often bridge many participating enterprises. Consumer Financial Services – Digital franchises, like AMZN, GOOGL, AAPL, FB, PYPL and others, will integrate services (e.g. payments, deposits, loans, etc.) from bank partners into their own services, dominating the customer relationship and commoditizing the underlying services. All in, we believe that initiatives in these areas could address more than $324B in annual banking costs, greatly reduce delinquencies, fraud and other risk exposures, and dramatically reduce the friction for financial transactions.

  • AI can enable better decisions and eliminate bottlenecks. Risk management – accurately assessing the likelihood of adverse outcomes that might affect the bank’s profits or capital and taking proactive steps to mitigate that likelihood – is at the core of the traditional banking business. Deep learning AI, able to draw subtle inferences from massive datasets and to react immediately to changing conditions, will have major impact on credit risks (liquidity, dilution, counterparty, concentration, etc.), and on operational risks (e.g. human errors, IT, fraud, security, etc.), with positive effect on exposures to market risks (e.g. hedging against exposures to interest rates, currencies, etc.). New AI infused processes could deliver better outcomes, while reducing costs and bureaucracy.
  • Blockchain enables low-friction, trustworthy transactions. Transactions of value require trust that promises made can be enforced – else, they will not be completed. Traditionally, that trust is established through intermediaries that add cost and time – clearinghouses, exchanges, inspectors, lawyers, escrow, various forms of insurance, government agencies, etc. Blockchain networks establish that trust through transparency. Each network member’s up-to-date ledger is an immutable and enforceable record of all members’ obligations and the historical provenance of all digital assets conveyed by the network. This has obvious applicability to almost all financial transactions – customer to customer, customer to bank, or bank to bank- making them less costly and more trustworthy with faster execution.
  • Digital platforms could front-end consumer financial services. Regulation restricts non-banks from delivering most financial services (deposit accounts, loans, etc.) directly, but digital platforms have begun to partner with banks to integrate them into their consumer franchises. AMZN, AAPL and PYPL are all reported to be launching credit products with their digital brands fronting for services provided by banking partners. Surveys of millennials suggest that they would be enthusiastic about buying most financial services (e.g. deposit accounts, credit, loans, etc.) through the top digital platforms – in particular, PYPL, AMZN, AAPL and GOOGL. This portends a disintermediation of banks from their customers, as digital partners leverage their consumer reach, data assets and cutting-edge analytics capabilities to control the customer relationship and commoditize the underlying banking service.
  • There are barriers. Regulation restricts the main elements of the banking business – loans, deposits, capital markets, etc. – to licensed banks, which are subject to comprehensive oversight of their capital, their operations, and their relationships with their customers. Furthermore, banking practices and relationships make it difficult for outsiders to gain the trust of consumers or traction with commercial customers. Many banking activities involve the cooperation of multiple banks and radical changes would require broad buy-in that will be difficult to achieve. This suggests that AI/blockchain fueled processes may take years to take root and that incursion by 3rd parties into consumer banking will be very slow.
  • Banks lack skills and infrastructure. AI talent is very expensive and highly concentrated amongst the top digital platform players – few banks even have one highly-cited AI scientist. Blockchain solutions require less esoteric skills, but also favor the participation of neutral parties to facilitate cooperation. Both technologies depend on high-performance, specialized datacenter infrastructure – banks are at a major disadvantage vs. digital platforms. Banks also lack reach to consumers, who increasingly shop on-line, do not often visit bank branches, and view banking apps with annoyance.
  • Three main opportunities for tech players. Tech opportunities in next generation banking will play out gradually over the next decade in 3 areas: 1. Risk Management Solutions – IT consulting companies (e.g. IBM, ACN, etc.) will help banks develop AI-based systems to automate credit, operational and market risk management processes, using tools and APIs from cloud hosting platforms (e.g. MSFT, IBM, AMZN, GOOGL). This will address $115B in costs for banks, in both commercial and consumer segments, while promising better outcomes. 2. Transaction Networks – Again, IT consultants and cloud hosts will be key to designing, implementing and operating blockchains that will typically need to be coordinated across many participating enterprises. This will address $209B in banking costs, while reducing friction that discourages transactions. 3. Consumer Banking – Digital platforms will integrate financial products (payments, deposits, loans, etc.) into their consumer franchises, cutting their banking partners’ direct relationships with consumers and commoditizing the underlying services. This would address $182B in profits earned by banks. AMZN, GOOGL, AAPL, FB, PYPL, and others will likely follow this tack, as much for its benefits in customer stickiness as its direct revenue potential.

The Brave New Bank.

Machine learning-based-AI allows computer programs to draw optimized conclusions from complex and seemingly ambiguous data – a perfect tool for the risk management at the core of most banking services. Blockchain enables participants in a network to have full confidence in the authenticity of assets being exchanged without investigation or guarantors – slashing the time and cost of financial transactions. Hyperscale datacenters in the cloud have costs several times lower than traditional private enterprise datacenters, while digital franchises engage hundreds of millions of consumers daily, with evidence that their users would happily turn to them for financial services.

These revolutionary technologies will have profound effects on banking, but there are clear obstacles. Banking regulation is comprehensive and strict, blocking unlicensed outsiders from offering even basic financial services and scrutinizing process changes within incumbents. Blockchains will require the close cooperation and investment of many banks – a difficult prospect in a competitive industry. Inertia will keep many banks and banking customers skeptical of new, automated processes.

The answer is a function of time and partnerships. Time, because regulators and customers will need it to absorb the implications of the new technologies and to facilitate change, and partnerships, because banks lack the skills and infrastructure to implement new solutions on their own. This will mean enormous opportunities for the platform-scale players and IT consultants leading the charge from the tech side. US banks reportedly spend $67B annually on IT – likely an underestimation, given fuzzy cost categories like power, real estate and people. At the very least, a good portion of that spending is up for grabs by cloud operators. Beyond that, the possible automation of risk management and transaction execution processes puts $250B of operating costs on the table.

Consumers’ banking relationships have changed dramatically from the days of passbook accounts and savings bonds. Surveys of millennials suggest that younger demographics are ready to ditch banks entirely, in favor of digital platforms – like PYPL, AMZN, AAPL, GOOGL, MSFT or even FB – because of their convenience and reputations for customer service. An incursion into the banking world is beginning – AMZN and PYPL are planning to offer deposit accounts and credit via bank partnerships, while AAPL is rumored to be working with GS on an Apple Pay branded credit card. This has the potential to go much further, with the digital brands fronting for bank-provided services. Tech companies could take from both ends – providing AI/blockchain/cloud solutions that cut the cost of consumer financial services, and then taking a margin on top of their increasingly commoditized bank partners.

Commercial banking is much less at risk of disintermediation. Still, early adopters that gain scale in their AI/blockchain solutions will have substantial advantage, and the IT consultants and hosting platforms that can enable that advantage will find enthusiastic demand. We believe IBM and ACN on the consulting side, and MSFT, IBM, AMZN and GOOGL on the hosting side, are best positioned to help design, implement and operate new processes based on AI and blockchain.

AI and Blockchain and Cloud – Oh My!

The emerging Cloud/AI era is already swallowing whole industries. Digital platforms capture 41% of measured media ad spending. Digital media has been eroding viewership on linear TV for 8 years, having already nearly killed recorded music purchases and print publications. Amazon will surpass Walmart as the west’s biggest retailer by gross merchandise volume before year end and is growing better than 30%. Cloud hosting on public hyperscale datacenter platforms is a $20B+ business, growing at nearly 50% per year and displacing investment in traditional IT. SaaS applications are capturing all growth in software spending, and then some – top names are growing at 20%+. Nearly 3 billion smartphones are in use world-wide, bringing the internet to many more people than anyone thought was possible. Self-driving robotaxis will soon begin picking up commercial riders in Phoenix, the beginnings of what we expect to be a global business worth hundreds of billions of dollars a year. Amazon, Berkshire Hathaway and JPMorgan have formed a partnership to look at using new technologies to cut the cost and improve the quality of healthcare for their employees.

Banking could be next. While a bulwark of regulation and inertia have thus far kept technology players from taking too much of a bite, we believe that a new wave of solutions will revolutionize the way banks operate – slashing costs, reducing time-consuming bureaucracy, improving outcomes, and reaching customers more conveniently (Exhibit 1). The technologies underlying these solutions – AI, blockchain and hyperscale datacenter operations – will come from the outside, and the banks that can partner to build early advantage and scale may be able leave their rivals behind. This is a huge opportunity for the companies able to provide that edge – companies like Microsoft, IBM, Accenture, Amazon, Alphabet and others.

The new tech is also creating an opening for popular consumer platforms like Amazon, Apple, Google and Facebook to exploit the engagement that users have with their services and front for traditional banking. Mobile payments – primarily Apple Pay and Google Pay – and on-line payments – like PayPal’s Venmo – are already changing the terms on credit cards. We think they could go a lot further than that, given the loyalty of their users, the depth of the behavioral data that they capture, and the strength of their data processing capabilities (Exhibit 2).

Exh 1: AI, Blockchain and Hyperscale bring Efficiency to Traditional Banking

Exh 2: The Key Areas of TMT-led Banking Disruption

Risk Management

Many of the internal processes at a bank serve to manage the financial risks that the organization bears – identifying each risk, quantifying it, and then mitigating it or hedging it as best possible. The Basel II Accords of 2004 organized banking risk into four categories with some overlap (Exhibit 3):

Credit Risk – Credit risks arise from the possibility that individuals or entities may fail to live up to promises made. Banks break these risks even more finely, with counterparty credit risk, dilution risk, liquidity risk, concentration risk, and others scrutinized to keep credit losses below thresholds. US loan delinquency, a measure of losses to default, are currently near historical lows below 2%, but spiked to nearly 7% during the depth of the great recession (Exhibit 4, 5). If a bank is too aggressive in providing credit, it risks higher defaults, but if it is too conservative, it rejects loans that are likely profitable.

Market Risk – In their activities, banks are exposed to changes in external market factors – interest rates, currency exchange rates, the price of commodities, the price of equity, etc. Banks are typically compensated for taking these risks on behalf of their customers but look to hedge their exposure as best possible. These risks are also amplified by the time span that the bank must have custody of the asset, so accelerating the pace of transactions is a strong mitigating factor.

Operational Risk – These are the risks that are internal to the bank’s operations. Human error and malfeasance fits in here – rogue traders and mistaken asset transfers still make headlines. Process failures – poor design can lead to bad outcomes – and IT risks – where automated processes fail – are also part of this, as are security/fraud risks. The cost of these risks is hard to quantify on a steady state basis, but obvious when failure occurs.

Residual Risk – Defined by Basel II as a bit of grab bag – everything other risk that a bank takes on in the job of being a bank. This includes geo-political risks, regulatory risks, and risks to the bank’s reputation.

The processes that work to control these risks require many employees working with very well-defined parameters to address every source of risk, from loan underwriting, to currency hedging, to fraud prevention. We believe that the cost of managing risk is roughly 11% of a bank’s expenses, working to protect 43% of its sales and almost all its assets. AI-based systems offer a better way – cheaper and more effective – to manage risk.

Exh 3: The Types of Risks Managed by Banks

AI Will Revolutionize Risk Management

Modern Artificial Intelligence systems use a growing roster of techniques using recursive algorithms that are loosely termed as Machine Learning (ML). We have written extensively on this (http://www.ssrllc.com/publication/a-deep-learning-primer-the-reality-may-exceed-the-hype/http://www.ssrllc.com/publication/ai-as-a-service-deep-learning-is-fundamental/http://www.ssrllc.com/publication/37440/http://www.ssrllc.com/publication/the-cloudai-era-a-perspective-on-the-next-decade-of-tmt-investing-2/). “Deep Learning” was coined to refer to particularly complex implementations of ML that employ these recursive algorithms in layers that have feedback mechanisms to capture data relationships at increasingly integrated levels. Altogether, these techniques allow computers to draw subtle inferences from massive databases that can be used to find optimal solutions to seemingly ambiguous problems. While the most well-publicized applications of AI have been in consumer applications, like the voice recognition in Amazon’s Alexa assistant, the image processing behind Google’s Photos app, or the autonomous driving systems operating Waymo’s robocabs, it will also drive a revolution in business analytics, enabling better-than-human performance in predicting outcomes and making decisions from new data.

Exh 4: Historical Quarterly Delinquency Rates by Loan Type, 1Q2013 – 1Q2018

Exh 5: US 2017 Annual Net Charge Offs by Loan Type for Commercial Banks