Funston Brief (Issue No. 24)
This month, our featured commentary discusses the CES 2024 show. We also dive into the recently released Lumiere model and showcase a very interesting conversation with Michael Ovitz.
After the text-to-image craze of 2022, AI video generation is poised to become a significant trend in 2024. Despite initial challenges and constraints, including copyright issues and short video duration, this technology is anticipated to significantly influence advertising, Hollywood, and other visual media industries. Its potential lies in facilitating cost-effective special effects and expediting post-production processes.
With its recent introduction of Lumiere, a space-time diffusion model designed for realistic video generation, Google has entered the mainstream text-to-video sector. Lumiere employs an innovative diffusion model named Space-Time-U-Net (STUNet), which effectively determines the spatial location of elements in a video and tracks their simultaneous movement and changes over time. According to Ars Technica, this approach enables Lumiere to generate videos seamlessly in a single process, eliminating the need to piece together smaller individual frames.
Lumiere initiates the video creation process by generating a foundational frame based on the provided prompt. Subsequently, it utilizes the STUNet framework to predict the movement of objects within that frame, producing additional frames that seamlessly flow into each other, creating the illusion of continuous motion. Notably, Lumiere surpasses Stable Video Diffusion by generating 80 frames instead of 25. Although Google has not been a prominent player in the text-to-video domain, its gradual release of advanced AI models and emphasis on multimodal capabilities, such as the Gemini large language model, indicates a strategic shift. While Lumiere is not currently available for testing, it underscores Google's potential to develop an AI video platform on par with (if not exceeding) existing generators like Runway and Pika.
ACTIONABLE INSIGHTS.
Edge AI from Google: The newest flagship phones from Samsung and Google’s own Pixel are the inaugural collaborators for the mobile editions of Google's Gemini generative AI model, a noteworthy development as it runs directly on the device rather than relying on cloud processing. Demonstrations showcase a range of features, including advanced photo editing and retouching tools, along with innovative concepts like auto-summaries for voice memos. This approach mirrors the pattern observed in the previous wave of AI, where novel AI features seamlessly transition into commonplace software as soon as they work.
Key Insight: A noteworthy addition is Google's innovative " circle to search " feature. This allows users to use their fingers to draw a circle around any item on their Android phone screen within any app, prompting Google to perform either a text or image search. For instance, circling a t-shirt in a video can lead to Google providing information on where to purchase it. Circle Press Release
Consumer Electronics Show: Steven Sinofsky writes a comprehensive annual roundup of CES2024. With a maturing supply chain for connected, low-power, inexpensive devices, CES2024 was the best in a long time. It is extremely clear from this year’s CES that we are seeing an “unbundling” of the smartphone. The reason is that the phone platform is stable at this point, and it is far too difficult for the incumbents to add something dramatically new. Link
The report showcases the highlights across the following dimensions (links take you directly to the respective sections in the report.)
EAR TO THE GROUND.
Podcast Episode: Invest like the best - Knowledge is power
Guests: Michael Ovitz
Takeaway:
Michael Ovitz is the co-founder of Creative Artists Agency (CAA.) This podcast covers the best stories and lessons from Michael's all-star cast of clients, why speed and scale were critical ingredients to CAA's success, and how he's translating his legendary work ethic and talent spotting to the investing arena.
“Scale is information, information is knowledge, and knowledge is power [...] CAA was a blend of American team sports, Asian philosophy, and just plain out-working everybody else. [...] I don't wait for things to cross my desk. I go out and look for them. I made 250 calls a day [...] The first meeting is easy. Anyone can get a first meeting. It's the fifth meeting, and how you're perceived that's critical. If there's no consistency and follow-up, you won’t build a business.”
- Michael Ovitz, co-founder of Creative Artists Agency
8%
In 2023, Microsoft, Amazon, Nvidia, and Google accounted for 8% of all North American VC dollars. The FTC is now reviewing this rise in financing, especially regarding AI startups. Link
PROFILE.
Jamie Dimon catapulted JPMorgan Chase to the position of the world's largest bank, amassing a staggering $51 billion in profits over the past year. Despite its pervasive presence in nearly half of American households through credit cards, loans, and banking services, Dimon's career is peculiar because he has never worked for a bank. When he assumed leadership at JPMorgan, the institution lacked a distinct personality comprised of various ancestors, cultures, systems, and businesses. Dimon, however, infused it with his own, shaping its unique identity. Enjoy the read!
“A business doesn’t have to be sexy to get him excited; it just needs to be reliable, profitable, and growing.” - Last Man Standing: The Ascent of Jamie Dimon and JPMorgan Chase
ON MY RADAR.
The Wager: A Tale of Shipwreck, Mutiny and Murder. This excellent book tells the true story of a British naval vessel shipwrecked off Patagonia's coast in 1740.