Flybridge AI Index: June 2024 Update
Summary:
In June, the Flybridge AI Index had a return of 7%. For the prior 12 months the Index had a return of 42% to its current level of 126% since January of 2023.
Of the 31 companies in the Index, 23 gained in June and 8 declined in the month. SIgnificant gainers include ARM (+29%), and Adobe(+26%). Significant decliners include Synapse (-7%), and Astera Labs (-5%).
The biggest news of the month was Apple's intelligence announcement and the confirmation of Apple's intention to be a leading player in the space.
We added three companies to the Flybridge AI index: Apple, Intel, and SuperMicro. The three companies are included in the Index from its inception and the Index's performance has been adjusted accordingly.
Apple announcement
Apple's June AI announcement was a major highlight, with key revelations about their AI strategy. Contrary to expectations of a proprietary large foundation model, Apple surprised many by announcing a partnership with OpenAI. However, they also unveiled an advanced personal intelligence system for their ecosystem, featuring an on-device 3-billion-parameter model. This model is fine-tuned for tasks like text generation, notification summarization, image creation, and streamlined app interactions.
This approach signals to founders that while large foundational models are valuable tools, they are not the entire solution. Apple is leveraging its strengths in on-device technology and user knowledge to enhance cross-application experiences within its ecosystem, and utilizing OpenAI models for more complex tasks. The expectation is that more model providers will be offered and that it won’t be an exclusive partnership.
Apple's announcement allowing third-party developers to access its new AI capabilities through the App Intents framework is significant. The keynote demonstration of Superhuman leveraging these AI features, despite Apple's native mail application, underscores Apple's strategy as a horizontal player. This approach offers solid foundational products across verticals while enabling specialized third-party players to create enhanced user experiences using Apple's AI capabilities.
An apt analogy is the automotive industry: Apple's mail application is akin to a Chevrolet or Toyota, sufficient for many users, while apps like Superhuman or Shortwave are comparable to BMW or Porsche, offering premium experiences that horizontal players struggle to match. This reinforces the importance for founders to create defensible, deep workflows through intimate customer understanding.
Base Layer the big winner
In the May edition, we highlighted how early AI era winners were primarily base/compute layer players. This month we quantified the difference: since the index's inception, the return of the base layer companies has been 159%, compared to 74% for the second and third layer companies (application or second-layer infrastructure like MongoDB). We expect this gap to narrow as more application layer companies go to production with their Gen AI features and start to see revenue generation and upsell from these products and features.
Interestingly, with the addition of Super Micro, NVIDIA is no longer the top performer. Since January 2023, Super Micro has returned 898% compared to NVIDIA's 746%. However, it's worth noting that Super Micro's LTM revenue was $12 billion, while NVIDIA's was $80 billion.
AI Insights and News:
Here are some noteworthy AI-related insights and news from companies within our index:
Qualcomm made its AI models available to app developers through the Qualcomm AI Hub. This announcement follows the introduction of its Snapdragon X Elite platforms, emphasizing the AI capabilities of the system-on-chips. (Source)
Cerebras, an AI specialized chip, has confidentially filed for an IPO. The company aims to capitalize on investor enthusiasm for AI hardware. Cerebras was founded in 2016, and raised over $720M to date. In march the company announced their CS-3, a cutting-edge, third-generation AI accelerator designed to train the most advanced AI models with unprecedented speed and scalability. It supports up to 1,200 terabytes of external memory, enabling the training of models up to 24 trillion parameters. (Source)
During COmputex AMD announced their AMD Ryzen AI 300 Series processors. These processors deliver exceptional performance across CPU, GPU, and NPU, with a focus on AI-driven experiences, offering up to 3X improvement in AI performance and up to 50 TOPs of AI processing power (Source)
At Computex 2024, Intel introduced new technologies focused on AI and power efficiency. Highlights included the launch of Intel Xeon 6 processors, Intel Gaudi 2 and 3 AI accelerator kits, and the Lunar Lake client processor architecture. Lunar Lake promises up to 40% lower power consumption and significantly enhanced AI and graphics performance. (Source)
Note: These insights are focused on AI developments and priorities as discussed in management publications, earnings calls, and other company announcements.
Performance Overview:
Inception (January 2023) to date returns for the Flybridge AI Index are 126%. In comparison, over the same time period the Cloud Bessemer Index returned 26%, the S&P 500 returned 42%, the Nasdaq returned 70%, and the F-Prime Fintech Index returned 82%.
The median NTM revenue multiple was 8.9x.
The median quarterly YoY revenue growth rate was 16%.
The median LTM net income margin was 13%.
Additions/Deletions:
We now welcome Apple, Intel and SuperMicro to the index, congratulations all for their inclusion:
Apple: People had been wondering and spreading rumors for months about what Apple's role in the new Gen AI wave was going to be. Last month, Apple finally unveiled their vision during their WWDC event. They shared their new intelligence focused on deeply personalizing user interactions with their devices while enhancing functionality through the use of advanced generative models. This new AI framework is integrated into iOS, iPadOS, and macOS to improve device capabilities in understanding and generating language and images, tailored to individual users.
Apple is integrating OpenAI's ChatGPT into Siri and systemwide Writing Tools, enabling users to access ChatGPT's extensive knowledge and generative capabilities directly within Apple devices.
Apple Intelligence uses context from user interactions to offer tailored responses and actions, making devices more intuitive and responsive.
Apple Intelligence will only be available on iPhone 15 Pro and iPhone 15 Pro Max models, as well as iPads and Macs with M1 chips or later
Intel: Despite the fact that so far people consider Intel a loser in the Gen AI wave, especially in comparison to NVIDIA, their announcement at Computex showed the priority they are giving to AI use cases, which fits criteria for inclusion in the Index. It will be interesting to see if Intel is able to catch up and start capturing some of the high market share NVIDIA holds for Gen AI applications.
Intel is concentrating on enhancing AI capabilities in cloud data centers with high-performance processors like the new Xeons, which offer increased cores and efficiency for AI-heavy workloads.
The development of AI accelerators like Gaudi2 and upcoming Gaudi3 chips focuses on optimizing AI workflows, providing an alternative to traditional GPUs for AI tasks. In their most recent earnings call, they announced that they expect over $500 million in accelerated revenue in the second half of 2024 from Gaudi 3.
Intel is pioneering AI integration into personal computers with AI-optimized chips and partnerships, aiming to make every personal computing device AI-capable.
SuperMicro: Providing advanced infrastructure solutions that cater to the demands of AI workloads, their extensive portfolio includes high-performance servers, cooling and storage solutions, and integrated systems designed for AI training, inference, and deployment across various industries. Supermicro collaborates with leading companies in the space, having been the first to bring optimized GPU platforms with NVIDIA in 2014.
Supermicro provides AI-ready edge computing solutions for real-time inferencing and predictive analytics, applicable in retail, manufacturing, healthcare, and smart spaces.
The company offers multi-tiered storage solutions that support AI training and inference workloads, capable of handling petascale data with high-performance bandwidth.
AI GPU platforms represented more than 50% of revenues in Q3.
AI-driven revenues in Q3 increased by 200% year-over-year.
SuperMicro offers direct-to-chip liquid cooling (DLC) solutions, emphasizing their ability to significantly reduce Opex and enhance efficiency in data centers. Implementing Super Micro's DLC solutions in a data center optimized for GPT-4 can save up to 30% on initial capital costs.
We also updated the methodology of the Index. To ensure consistency, we will only include companies listed on U.S. exchanges. For this reason, Samsung and SK, which are listed outside the U.S., are removed from potential additions.
Another modification is that from now on, when a new company IPOs that fits the criteria like Tempus AI, we will add them three months after their listing. The reason is that volatility in the months following an IPO tends to be very high.
For compute-related companies, we established a threshold to include only those directly involved in first-tier computing technologies. We exclude companies further removed from core computing, such as those producing semiconductor fabrication machinery (e.g., ASML) or raw materials suppliers. This boundary ensures our index focuses on companies with closer proximity to the application layer. This approach aligns with our goal of providing relevant insights for startup founders in the AI space.
