Summer in Review ☀️😎
Capital Markets Update, Llama-2, Nvidia, and the 2023 Intelligent Apps Summit!
Thank you to our now 1.5K+ subscribers, we appreciate all the continued support! Please continue to share with colleagues and friends :)
We hope you are all enjoying the remaining weeks of summer! Can you believe it’s nearly Labor Day Weekend?!
A lot has happened since the start of the summer, so we thought it would be worthwhile reflecting on the last few months and highlighting a few of the big events that have occurred. We’ll also take a quick look forward to the rest of the year and highlight Madrona’s upcoming Intelligent Applications Summit!
Here’s what we’ll dig into:
Capital Markets Update in AI (IPO, Strategic M&A, and PE Activity)
Noteworthy AI Research, Important AI News, & Great Summer Reads
What to Look Out for the Remainder of the Year (IA 2023 Summit)
We hope you have some fun plans for the Labor Day Weekend, so we are keeping this week’s post as a shortened version, but we’ll be back in the full swing in 2 weeks - hope you enjoy!
Capital Markets Updates in AI (IPO, Strategic M&A, PE):
IPO Updates: There’s no better way to end the summer than knowing that the IPO window is cracking open, which was propelled by Klaviyo filing their S-1 last week. This is a huge milestone, as this is one of the first filed software IPOs we’ve seen in almost 2 years! Just for context, there were 2 Technology IPOs in 2021, 124 in 2021, 58 in 2020, and 42 in 2019. Klaviyo’s S-1 filing begs the question - is this the start of the re-opening of the IPO markets, particularly for SaaS companies?
Klaviyo Files for IPO - Klaviyo, a customer data platform that ingests marketing data and powers customers to automate their marketing with data and AI files their S-1. We recommend reading Jamin Ball's analysis of benchmarking Klaviyo’s S-1 Data.
Strategic M&A Updates: While there has been a general pullback in software M&A, we were heartened to see a good volume of activity in the applied AI/ML space over the summer. While the GenAI space is still in its infancy, we have seen M&A occur at both the infrastructure and application layers. This reinforces our thesis that many “Gen-Enhanced” companies are looking for ways to become more competitive against their peers by offering more “Gen-Native” capabilities. We expect there will continue to be more private-to-private M&A. A select number of these include:
DataBricks Acquires MosaicML: June 2023 for $1.3B.
Thomas Reuters Acquires CaseText: August 2023 for $650M Cash
Digital Ocean Acquires PaperSpace: July 2023 rumored for $111M
Thoughtspot Acquires Mode: June 2023 for $200M
Financial Sponsor (Private Equity) Updates: Since the early 2000s there has been an explosion of software-focused Private Equity firms (e.g., SilverLake, Francisco Partners, Vista Equity, Thoma Bravo, etc.). These financial sponsors are sitting on billions of dollars of dry powder and we believe they will be major catalysts in acquiring VC-backed software unicorns that are facing liquidity crunches. Depending on multiples, we may also see many public companies taken private. It will be interesting to watch what the “AI strategy” will be for these Sponsor-Backed platforms, as historically many Sponsors will acquire one platform and perform bolt-on acquisitions over time.
New Relic Acquired (Take-Private) by Francisco Partners & TPG: July 2023; $6.5B Equity Valuation, representing ~26% to New Relic’s 30-day volume-weighted average closing price ending on July 28, 2023, and ~30% premium to New Relic’s last-twelve-months volume-weighted average closing price ending on July 28, 2023.
Noteworthy AI Updates & Interesting Reads:
There are many great newsletters that summarize what’s going on in AI on a weekly basis (we recommend Ben’s Bites and Through the Noise), but below we’re highlighting a handful of the interesting summer updates across business models, LLMs, and new application ideas that we think will be influential in the coming months ahead.
Business Updates:
Nvidia - Nvidia, the world’s leader in AI computing manufacturing, announced FYQ2 earnings and delivered against heightened investor expectations. Growing investments in GenAI infrastructure drove a 21% revenue beat and dependence on NVIDIA GPUs has become one of the roadblocks to accelerating innovation in the AI space. We recommend reading Ben Thompson’s piece “Nvidia on the Mountaintop” and more on the GPU supply and demand/shortage debate here.
Microsoft - Microsoft announced Microsoft 365 Copilot earlier this year and in July they made the decision to make it available at the relatively steep price of $30/user/month. This premium is noteworthy considering Microsoft currently charges businesses $36/user/month for Microsoft 365 E3, which includes access to Office apps, Teams, SharePoint, OneDrive, and other productivity features. We are curious how the market will react toward this pricing model. In general, we’ve found pricing to be a tricky topic for many gen-native application companies.
OpenAI - News leaked recently that OpenAI is on track to generate $1 Billion in annual revenue (that’s billion with a “B”), an eye-popping number relative to their past annual revenue of <$50M. This stands in contrast to reports we’ve seen and talked about where ChatGPT’s user count has been flattening. This implies that both enterprises and consumers are still finding tremendous value in premium ChatGPT plans, and OpenAI is entering elite territory in the very few companies generating over $1B in revenue.
Model Updates:
Llama-2 - Meta and MSFT Introduce the Next-Generation of Llama, Llama 2. Llama 2 is another indication of how robust innovation is happening in open-source, particularly in making these LLMs accessible for everybody including startups, researchers, and businesses. Armand Ruiz talks about why the Llama 2 launch is a Game Changer. We encourage giving this post a read!
PaLM 2 - Google announced a whole slew of updates this week at Google I/O, including several updates to their proprietary LLM PaLM which has four times more input token length and is available in 38 languages! You can read more on their announcements around PaLM 2 and Vertex AI from the conference here. We view this as an important milestone, as Google is still set on a path of trying to beat OpenAI and ChatGPT and the race is on!
App Ideas that Gained Traction this Summer and We’re Keeping a Close Eye On:
A Personalized Assistant for You - What if you could have a personalized and private agent that understands everything about you? Companies like Gather, and Inflection (which announced a $1.3B Investment Round in July 2023) are tackling this interesting problem. We’re excited to see the continued innovation in this space.
The New Frontier for LLMs is in Biology - While AI and LLM-powered protein design is still nascent, we are excited about how LLMs can create new protein designs. New novel proteins may serve as therapeutics for a wide range of human illnesses, from infectious diseases to cancer, and so much more that we cannot yet even imagine.
Worthwhile Reads From the Summer:
An Interview with Daniel Gross and Nat Friedman about the AI Hype Cycle (Beyond a Paywall, but well worth a read/listen)
Generative AI Strategy by Chip Huyen: If you are looking for a simple framework to explore what to do with Generative AI, we highly recommend this read by Chip Huyen, who also has many other great posts and a book on Designing Machine Learning Systems!
Patterns for Building LLM-based Systems & Products by Eugene Yan: If you are interested in learning about practical patterns for integrating LLMs into systems & products, this is a worthwhile read by Eugene Yan!
Looking Ahead To the Rest of The Year:
As we look toward the remainder of the year, we’re excited to understand how the AI landscape will continue evolving. We (at Madrona Ventures) are also particularly excited to be hosting our 2nd Annual Intelligent (and Generative) Applications Summit on October 10th and October 11th in Seattle! We will be bringing together ~300 leaders, researchers, and executives building in the AI/ML space.
Below are 5 key topics we’ll be exploring at the summit. Many of these questions also remain top of mind for us as we think about the remainder of the year:
What are the best practices for building gen-enhanced and gen-native applications? There are a growing number of approaches to enhancing existing apps with GenAI capabilities and there are a whole host of new companies who are also emerging as Gen-Native companies. Who wins in the short term, vs. who wins in the long term, and are there spaces where Gen-Native is better?
Will one model rule them all, or will many models to rule them all? What is this notion of model-cocktails? How do you leverage the best of different types of models for specific tasks or model outputs?
What are the key emerging architectures for building generative applications? Why is Retrieval-augmented generation (RAG) a critical component of the AI framework?
How do you extract data from different systems and different formats to ultimately create more personalization of model outputs?
What are the tradeoffs between domain-specific and horizontally applicable models within GenAI? How does this apply across different verticals like healthcare, finance, CPG, industrials, etc.?
Of course, much will change between now and the end of the year…so, if you have any interesting questions you’d like to raise, or are interested in joining in on the conversation, please DM us or apply to attend the summit here.
We’ll leave you with one of our favorite keynotes from the 2022 Summit - enjoy!
Thanks for joining in on our shortened version of Aspiring for Intelligence! We hope you have had a great summer and enjoy the long weekend. We will see you again in two weeks!
This is my first reading of this newsletter. Found it informative and helpful. Some great references to interesting reads on GenAI, how models are evolving and other helpful perspectives in this space.