Microsoft 3Q23 Earnings Highlights
Revenue increased 7% YoY to US$52.9bn from US$49.4bn. Intelligent Cloud revenue grew the most at 16.3% YoY to US$22.1bn due to the strong performance of Azure services which grew 27%.
Operating income was $22.4 billion and increased 10% YoY from US$20.4bn.
Net income was $18.3 billion and increased 9% YoY from US$16.7bn.
Diluted EPS was $2.45 compared to $2.22 in the previous year.
Other than financials, Microsoft revealed how rapidly OpenAI is growing with a 10x growth QoQ in the number of customers, reaching 2,500 customers utilizing the Azure OpenAI Service. GitHub Copilot, Microsoft's AI developer tool, also saw massive adoption reaching 10,000 customers onboarded on the Copilot for Business service. And it seems that OpenAI has really made Bing a feasible competitor to Google Search with Bing reaching 100 million daily active users just two months after the launch of the new Bing and Edge. Daily installs have also grown 4x since launch.
For the other segments, management guided that Azure is likely to be able to sustain its 27% growth rate into the rest of the year, owing largely to AI services. Microsoft's gaming segment continues to underperform, especially Xbox hardware revenue, which fell 30% YoY. To note, Microsoft's acquisition of Activision Blizzard has been blocked in the UK, dampening prospects of the company boosting gaming revenue later in the year.
Microsoft Acquisition of Activision Blizzard Faces Headwinds
While management did not provide much guidance on this issue, the blockage of Microsoft's acquisition of Activision Blizzard comes as a great point of contention for many as some view the gaming market as a market filled with alternatives to the major IPs held by both Microsoft and Blizzard such as Call of Duty and World of Warcraft, meaning the acquisition shouldn't be any antitrust concerns. However, Britain's CMA and US's FTC both stated that the cloud gaming sector was the main concern.
The cloud gaming sector is famous for being a graveyard of many cloud gaming platforms such as Stadia, OnLive (acquired by Sony), Gaikai (acquired by Sony), and Ouya (acquired by Razer). But there are more platforms that are available right now than what the FTC and CMA seem to think. The most popular cloud gaming platform now is Nvidia's GeForce NOW which allows you to play PC titles even on your mobile phone. The other platforms are Microsoft's Xbox Cloud Gaming and Amazon Luna. While the acquisition will give Microsoft control over Warcraft, Diablo, and Overwatch IPs, these games have lost a significant chunk of their MAUs owing to the various controversies at Blizzard which have slowed game development for these IPs.
The Cost of Advancing Generative AIs
During the earnings call, management received various questions asking about the cost of AI compute workloads and if it will become an increasing cost concern, especially on the OpenAI API side of things, but if you think about it, each workload is valuable data on which OpenAI is utilizing to rapidly develop their foundation models. Given the early stage of the industry, the development benefits provided by massive AI workloads outweigh the costs. OpenAI and Microsoft understand the importance of keeping their lead in the AI market, which is something you can't put a price tag on.
Google 1Q23 Earnings Highlights
Revenue increased 3% YoY to US$69.8bn from US$68.0bn last year. While revenue growth was lackluster, the Google Cloud segment showed the greatest growth of 28.1% YoY to US$7.5bn, driven by increases in both seats and average revenue per seat.
Operating income fell 13.3% YoY to US$17.4bn with margins shrinking from 30% last year to 25% this year primarily due to workforce and office space reductions.
The Google Cloud segment has finally turned profitable after 3 years of losses with an operating income of US$191m.
Net income fell 8.4% YoY to US$15.1bn from US$16.4bn last year.
Diluted EPS was $1.17 per share compared to $1.23 per share last year.
Beyond financial highlights, Google's management delivered a very AI-heavy earnings call, indicating that Google is very determined to take a cut out of OpenAI's cake. Google's CEO himself hinted at the company's focus on integrating its generative AI capabilities into the rest of the Google suite, especially Search.
The most significant announcement was that they were combining the Brain Team in Google Research and DeepMind, Google's AI and ML team, to accelerate the development of their AI projects. It is likely that one of the initial priorities of this new team would be the implementation of Google's PaLM API, its API to access Google's LLMs, across the rest of the product suite. Management also guided that Google has completed its layoffs and is now stepping up investments in both data center construction and servers from 2Q23 to the end of the year.
Google had a Lead in AI but Lost it
Google had delved into the AI space as far back as 2006 with its first acquisition of Transformic, a company specializing in machine learning and data mining. The company had at least 11 more AI-related acquisitions up to 2021 including DeepMind which was acquired at US$500m back in 2013. For perspective, Microsoft and Amazon only acquired their first AI company in 2015 and 2013 respectively. Although all three hyperscalers made similar amounts of AI-related acquisitions, DeepMind was the highest-profile deal of that period in the AI industry. Despite this, DeepMind and Google couldn't develop a foundation model that could compete with OpenAI. It begs the question if the merger of Google Research and DeepMind will yield the expected AI improvements Google has been touting.
Amazon 1Q23 Earnings Highlights
Revenue increased 9% YoY to US$127.4bn from US$116.4 last year. AWS segment revenue increased 16% YoY to US$21.4bn, weaker than expected as customers optimized cloud spending given the tough macroeconomic situation.
Operating income increased 29.7% YoY to US$4.8bn from US$3.7 last year.
Net income was US$3.2bn compared to a net loss of US$3.8bn last year.
Diluted EPS was $0.31 per share compared to -$0.38 per share last year.
Out of the three hyperscalers, Amazon was the only one that keep the AI announcements light and focused more on its cloud business. Management stated that they are focusing on retention and assisting customers with optimizations for AWS due to macroeconomic headwinds, however, its new customer and migration pipeline remains strong.
AI Could be the Sector that Makes or Breaks AWS
Management has stated that CAPEX is being diverted from fulfillment and transportation to AWS for LLMs and generative AI even though overall CAPEX remains roughly the same from last year. When prompted with a question on how AWS compared against competitors, management defended by stating that it isn't just developing foundation models but improving its own AI chips called Trainium, which specializes in inference and predictions. Which is almost no differentiation considering that each hyperscaler has its own AI chip or is developing one.
After this recent earnings, its no news that AWS is losing cloud market share to Azure and GCP. And this losing streak could potentially be extended further with the onset of generative AI, which as you know, Microsoft and OpenAI are leading. If AWS's foundation model bet does not perform at a level close to OpenAI's models, chances are companies will look to Azure cloud services to future-proof themselves for new OpenAI integrations and services that Azure may offer down the line. Arguably, Google may also be ahead of Amazon in terms of generative AI. With the launch of Bard and PaLM API, Google also has been a step ahead in gathering data from users to refine its foundation models. It can be said that AWS's future depends on the performance of its foundation models and AI division.
Comments