☀️☕ Google this: "What is the Innovator's Dilemma?"

30 January 2023

☀️ Happy Monday!

📝 Focus

• Google this: “What is the Innovator’s Dilemma?”

📊 In the Markets

• AmEx: Gen-Z don’t leave home without it

• Adani Group companies plummet

• Intel In... a hole?

📝 Focus

Google this: “What is the Innovator’s Dilemma?”

And then chuck the same question into OpenAI's ChatGPT:

Giving direct answers to queries, rather than simply pointing users to page after page of ranked, suggested links, would likely result in fewer searches. Is Alphabet (Google), currently the fourth most valuable company in the world both by earnings and market capitalisation (number of shares X share price) and its hugely profitable advertising business model built around search, facing a classic "Innovator’s Dilemma" 🎓?

Google already has DeepMind, acquired in 2014, and its in-house Google Brain deep in the Artificial Intelligence (AI) space. Its Pathways Language Model (PaLM), released last year, can be scaled up to 540 billion parameters, three times that of GPT-3, with a mere 175 billion. (Performance across tasks increases with the model’s scale.) Its NLP algorithm BERT has been powering Google search for years, enabling them to migrate from keyword-based to contextual-based results and grouping several related news articles together in carousels. And Google is actually already using its GPT-3 conversational chatbot, Language Model for Dialog Applications (LaMDA) in its Google Chat function, where it summarises an ongoing conversation in the chat window (with the approximately 2 people in the world who know Google Chat exists.)

Besides fewer searches, leading to lower advertising revenue, Morgan Stanley Research estimates that answering a search query using AI language processing API calls would cost seven times as much as a standard search.

One of the biggest constraints on Google’s broader release of AI tools like GPT-3 has been reputational risk. Google points out that Generative AI always runs the risk of incorrect or inappropriate results, and could exacerbate the polarising effect of social media. In 2016, Microsoft launched and quickly shut down "Tay", its artificial intelligence chatbot designed to develop conversational understanding by interacting and learning from humans on Twitter. Twitter, yes. You can already see where this is going. Microsoft set up Tay with a young, female persona to appeal to Millennials, and within 24 hours, Twitter users managed to get Tay to post things like "Hitler was right I hate the Jews" and "Ted Cruz is the Cuban Hitler."

Meanwhile, Apple is using its iOS, with its lock on 1.2 billion higher-end consumers, to launch its own assault on Google search, even though it receives billions a year from Alphabet to provide Google as the default search engine on iPhones. Fabled co-founder Steve Jobs did call Android “a stolen product”, after all, and declared 2011 the year of "holy war against Google."

It's already using a feature known internally as “Apple Search”, with “billions of searches” per day, according to the FT, when iPhone users ask Siri for information or do searches from the home screen, or when users use the “Spotlight” search feature on a Mac. The purpose of attacking the search space from Apple's perspective is to get huge amounts of first-party data to power its move further into the digital advertising space that Google (and Facebook parent Meta) still dominates. Its customer data and privacy-centric approach has worked well in the current environment and could be applied to its search offering, too.

On the other hand:

📊 In the Markets

Investors took their time to react to most of the results that started to stream in last week given that the Federal Reserve has its upcoming interest rate decision on Wednesday, where a 0.25% hike is almost universally expected (99.2% implied by futures market pricing on the CME.)

Supporting this was the Fed's preferred inflation gauge, the core personal consumption expenditures price index (excludes energy and food) which showed prices rose 4.4% from a year earlier, the lowest in more than a year, from 4.7% in November. On a month-on-month basis, December prices were up by 0.3% (as expected) compared to a 0.2% rise in November. After (broadly) better-than-expected economic growth for 4Q the day before, the chances of a "soft-landing" or at least only a mild recession are encouraging, despite data showing that real consumer spending fell 0.3% in December following a 0.2% drop the previous month.

Shares of Tesla had another double-digit day, though, surging 11% and adding US$55 billion to its market value (number of shares X share price) in a day.

And credit card company American Express, a Dow Jones Index component, added US$12 billion to its market value by jumping 10.5% after reporting a 25% increase in full-year revenue to a new record.

AmEx: Gen-Z don’t leave home without it

Founded by the same Henry Wells and William Fargo who later set up Wells Fargo, American Express was originally an express shipment dispatcher (goods, securities, currency, etc.) but is now a globally integrated payments company most visibly with its charge and credit cards. Its largest shareholder with 20.3% is Warren Buffett's Berkshire Hathaway, where AXP is also one of its largest holdings (with the first ​​5% of the company bought for US$20 million in the 1960s.)

American Express (AXP) missed analyst forecasts as its fourth-quarter profits fell by 9% as it set aside much more than expected to cover potentially bad loans, US$1.03 billion compared to only US$53 million in the same period a year ago, while writing off 1.3% of its total loans, compared to only 0.8%.

But what the market really liked was that it raised its quarterly dividend and forecasted higher-than-expected profits for 2023.

Though the company saw bad loans rise (even among its generally better-off and creditworthy customer base), AmEx said that the company’s credit metrics “remained strong” and that it is seeing few signs of a recession in the short-to-medium term, noting that cardmember spending continues to remain strong.

During the pandemic, it leveraged its perception as a status symbol and took a bold and successful move to get new, high-spending cardmembers with an even more expensive Platinum tier card loaded with better rewards and access to TikTok-friendly luxury Centurion airport lounges. It doubled the number of Platinum cardholders, with millennials and Gen Z customers making up 60% of all new cardholder growth.

AmEx has spent years pivoting its traditional charge card business model, where a customer pays off its entire balance each month, to a model more similar to that of a traditional credit card issuer. Instead of just relying on membership and merchant fees, AmEx would then "encourage" customers to keep rolling over a balance so that the company can charge interest off of that balance as well as late fees.

Rivals Visa and Mastercard sit as payment processors between bank card issuers and merchants to facilitate payments. They get a small share of the “interchange” fee for intermediating this process, which involves routing data between a consumer and the merchants' banks and getting the bank to settle funds net of fees.

American Express does much the same with banking partners but also issues cards under its own name and lends from its own balance sheet while charging merchants more to use its payment processing network (in order to access AmEx's higher spending cardmembers). Worldwide "cardmember" loans were $108 billion last quarter, up 22% from 2021, while for the full year, they spent more than US$1.5 trillion on their cards last year, a 21% jump over 2021.

Adani Group companies plummet

Meanwhile, in Indian markets, the Hindenburg Research short call on Adani has led the sell-off to hit US$52 billion, with flagship company Adani Enterprises plunging 18% on Friday, dragging down the BSE SENSEX, which fell 1.45%. The wide public release of the research report was clearly timed to appear just days before Adani Enterprises' share sale to raise Rs200bn (US$2.4 billion) from global investors. The first day of retail bidding on Friday only saw 1% of the offer subscribed, with the price now 11% below the INR3,112 minimum offer price, and bankers are reported to be considering extending the Tuesday subscription closing date by four days.

Intel In... a hole?

Semiconductor giant Intel, once a leading light in US tech, saw US$8 billion wiped off its market value on Friday after dismal results, with sales down 32% and a loss of US$661 million from a profit of US$4.6 billion a year ago. Next quarter is expected to be worse.

In particular, the chipmaker showed how badly it was doing in the crucial data centre business against long-time rival Advanced Micro Devices, even as it flounders in its head-to-head conflict for contract manufacturing business against behemoth Taiwan Semiconductor (TSMC.)

Not coincidentally, AMD is a fabless (i.e. design only) chipmaker which uses TSMC technology that trounces Intel's, and as a result, offers chips that have a strong price-performance advantage over Intel's equivalent products. AMD used to be a fraction of Intel's size, but since spinning off its manufacturing business (as GlobalFoundries) in 2009, has caught up and exceeded its old rival, under Taiwanese-American CEO Lisa Su (since 2014.)

Intel is also facing off against another fabless chipmaker customer of TSMC's, Nvidia (currently Wall Street's hot pick as a play on Artificial Intelligence -- like ChatGPT.) Again the largest semiconductor company by market value, Nvidia (under Taiwanese-American CEO and founder Jensen Huang) is beating Intel not only in graphics processors but may also be moving into Intel's backyard with central processing units. All this while former Intel CPU customers, including Apple and Amazon, are designing their own chips.

Mini-Explainer: Semiconductor Companies… Fabless and Foundries

• Semiconductors are the basic building blocks of an ever-widening range of electronic products, far beyond just computers and electronic items. The word's often used interchangeably with "microchips" or just "chips."

• Semiconductor companies may have their own factories (fabrication plants, or "fabs") or not ("fabless", where they just do the design, such as Qualcomm, Broadcom, Nvidia... and Apple.)

• A fabless semiconductor design house needs its chips built by a contract manufacturer or "foundry" which invests enormous sums in its plant, equipment and, more importantly, people and processes. TSMC is by far the largest, most advanced and most profitable foundry in the world.

In collaboration with Micro Digest 📚

Subscribe here!