Contrary Research Rundown #81
AI's Cost Conundrum, plus new memos on Chainalysis, Ada, and more
On May 22nd, we're hosting our next Tech Talk with Ramp, Hex, Sourcegraph, Semgrep, Onehouse, and Weaviate in San Francisco. The evening will feature eng demos and conversations with senior leaders from the featured companies, followed by food and drinks. Register for the chance to join!
Research Rundown
The cost components of AI continue to rear its ugly head. Building on the story of Stability’s CEO departure, it was recently reported that, while the company had generated ~$11 million of revenue in 2023, its cloud computing bill was $99 million. The company had attempted to raise more capital to bridge the gap, but had been unable to raise enough.
While the cost of compute can massively drag down a business, its also clear that identifying a use case for a breakout AI product does not immediately follow hype and excitement. AI-generated presentation startup, Tome, has reportedly struggled to find the ideal use case for its product, between enterprise and consumers. In addition, the business is dependent on model providers, using ”OpenAI and Anthropic’s models for its text generation, and Stable Diffusion for image generation.”
One investor quoted the Databricks team as articulating something they call “Mosaic’s Law,” stating that “the cost to run AI models will decrease by 75% annually due to hardware and algorithmic improvements.” While that may eventually be true, it feels optimistic given the fact that OpenAI spent ~$4.6 million training GPT-3, compared to spending $100 million training GPT-4.
Even Elon Musk, who is reportedly in talks to raise $4 billion at an $18 billion valuation for xAI, emphasized in his pitch that “xAI will be able to train on the high-quality data from Musk’s social network X.” So who knows, maybe spending $44 billion on a social media platform to generate your own training data will end up being the most economical way to build AI models.
One thing is clear. The race is on between AI application companies rushing to identify lucrative use cases for their products, while AI infrastructure companies hope to take advantage of “Mosaic’s Law” in dramatically reducing the costs behind AI.
Ada is a company that offers an AI-powered platform designed to automate customer service interactions in order to improve customer satisfaction and reduce the workload on human customer service professionals. To learn more, read our full memo here and check out some open roles below:
Engineering Manager, Channels and Handoffs - Remote Canada
Chainalysis is a blockchain data platform that provides intelligence, risk, and growth solutions and services to government agencies, financial institutions, cryptocurrency businesses, and consumer brands. To learn more, read our full memo here and check out some open roles below:
Software Engineer II, Global Search - Mexico City
Account Executive, National Security - D.C. Office
Check out some standout roles from this week.
Traba | United States, Canada - Software Engineer, Staff Software Engineer (Frontend), Senior Software Engineer (Backend)
The Financial Times published a piece called “How Google lost ground in the AI race,” outlining how the company’s response to ChatGPT and the launch of products like Gemini was mired in organizational and strategic misdirection
As robotics become increasingly important amidst an “aging population and declining birth rates,” China leads the charge in installing industrial robots. In 2022, China installed 6x more industrial robots than the U.S.
One of the limiters for robotic installation is fears about job destruction. Stefano La Rovere, Amazon’s director of global robotics, believes the opposite. “Over the last years, more than 700 new categories of jobs have been created by the use of technology.”
In a new batch of Arena results, Cohere’s open model, Command R+, has shot to sixth place, both matching results from models like GPT-4-0314, and being the highest ranked open model.
Former GitHub CEO, Nat Friedman, posted an interesting chart showing web traffic for AI products Claude, Gemini, ChatGPT, and Bard. Two takeaways: (1) Gemini traffic is already 25% of ChatGPT’s traffic, despite Google not yet pushing it through large distribution channels, and (2) ChatGPT’s traffic has been relatively stagnant.
Nick Shalek, a General Partner at Ribbit Capital, made an investment pitch for Ethereum in front of a group of institutional investors at the Sohn Conference.
Increasingly, large tech companies are starting to invest more in custom chips or similar operations. First, there was Sam Altman’s pitch for Microsoft and OpenAI to work together on a chip project, meanwhile Microsoft announced two custom chips last November. Now, Meta has announced custom chips for its AI workloads.
At Contrary Research, our vision is to become the best starting place to understand private tech companies. We can't do it alone, nor would we want to. We focus on bringing together a variety of different perspectives.
That's why we're opening applications for our Research Fellowship. In the past, we've worked with software engineers, product managers, investors, and more. If you're interested in researching and writing about tech companies, apply here!