Posts

I remember my graduate classes well : marketing, operations research, statistics, macroeconomics, strategy, & negotiation.

But having worked for about twenty years, there’s a class missing from business school for a skill used as frequently as public speaking & negotiation : Delegation.

After starting in the working world, at some point we start managing others. Perhaps just for a project. Perhaps a small team. Or an entire business unit. Or working with an executive assistant.

Open source models have become a critical part of the AI landscape.

I was curious about the trends in the open source ecosystem, so I analyzed HuggingFace data on the top 300 open source models, both by overall usage & also the top of the trending list.

Open source models are governed by open source licenses. Similar to regular open source software, Apache & MIT dominate the licenses by model count. 76% of the top models choose one of these licenses. Apache is nearly twice as popular as MIT.

The 10x programmer - engineers whose prodigious productivity and perspicacity change the arc of a business.

But what about a 10x AE? A salesperson who books 10x more business than their next closest peer.

10x engineers use their knowledge of technology to achieve astounding feats. Why wouldn’t the same idea apply to sales?

With AI, we will start to see sales outliers who meaningfully outperform & break quota expectations.

AI compresses prospect research time. AI writes emails. AI captures transcripts from sales conversations. All of this work is 50-80% automated.

Measuring engineering productivity is a Gordian knot.

How does a manager measure the performance of an engineer? Lines of code, like lines of a blog post aren’t a good metric. Most of the time, short, direct prose is better than verbose or lengthy sentences that carry on forever because they haven’t been edited & really ought to be, but someone was rushing or forgetful, & they lose the reader along the way.

Companies aren’t bought. They’re sold.

Hang around startups & venture capital long enough, & you’ll hear this aphorism. It means it’s much easier to sell a company when there’s plenty of demand.

In any given year, there’s an 82% chance the value of startup liquidity will change by more than 77% - both up & down.

Aside from certain crypto tokens, it’s hard to find a less reliable market than the VC exit market.

Every year I make a list of predictions & score last year’s predictions.

Here are my predictions for 2024.

1. The IPO market remains closed through the first 6 months of the year. But a few mega issuances, especially Stripe & Databricks in the summer or fall, re-open it for others. The Fed cuts rates, which helps.
2. M&A accelerates throughout the year. The anticipation of a rate change drives fear of target acquisition valuations. In the last two years, M&A has totaled about $49b & it surges to above $60b driven by AI acquisitions. PE becomes an important buyer of companies growing 10-25%, as it did in 2018, driven by lower debt costs.
3. AI & data continue to dominate the funding landscape as founders & investors seek novel applications of the technology. A handful of companies achieve record-setting growth rates.
4. The share of AI-enabled web searches approaches 50% of all consumer search as consumer behavior patterns evolve, especially on mobile.
5. The BTC ETF drives a resurgence in interest in web3 financing. The winter forced many companies to evolve from open-source projects to revenue-generating businesses. We see the first broadly successful tokens with dividends (likely outside the US). This innovation reinvigorates very early-stage IPOs. We also see more ARR-based web3 businesses achieving scale. Record inflows into tokens fuel all-time highs in Bitcoin, Solana, & higher performance L1s who offer better price/performance to market.
6. US VC investment falls from $275b in 2022 to $200b in 2023 & sustains at about $200-220b in 2024 as LP interest in venture attenuates after the euphoria in 2020 & 2021. Valuations remain relatively steady except for AI businesses, which command a premium to market of about 10-25%.
7. The discussion around AI regulation becomes a critical topic in the US during the election because machine-generated content exacerbates international meddling in US politics. But the overwhelming desire for the US to continue to lead the innovation wave it started creates safe harbors, the same provisions which enabled the web to flourish, are applied to AI.
8. Companies & startups in particular begin to report meaningful improvements in productivity from AI, reducing their headcount growth, but growing revenue just as much as projected. ARR per employee increases 10%, twice the decade long average.
9. Data lakes become the dominant data architecture across business intelligence & observability workloads as more startups leverage Amazon S3 free replication. Cloudflare R2’s architecture for very large data sets drives a meaningful growth in its usage, predominantly for AI.

Grading last year’s predictions :

Earlier this week, Alteryx announced its intention to go private for $4.4b. It’s the 10th in 2023.

In December of last year, I wrote :

Private equity acquires 10% of the 70+ publicly traded software companies by the end of the year. With hundreds of billions of dry powder, plenty of healthy cash flows generated by SaaS publics, & the leverage of the inevitable shareholder lawsuit if a board rebuffs the 30% premium of a tender offer, private equity becomes the dominant M&A option in dollar terms for 2023.

Data teams are becoming software engineering teams.

On December 14th we welcomed Philip Zelitchenko, VP of Data from ZoomInfo, to talk about how he has built this discipline within his team & it was fascinating. The video is here.

Like the Devops movement, the Dataops movement aims to scale the use of data within companies without increasing the headcount of the data team.

I had pre-ordered Poor Charlie’s Almanack before Charlie Munger died & read it when it arrived.

In the ten speeches that constitute the book, Mr. Munger reveals his frameworks for life.

Mr. Munger was known for his worldly wisdom, a collection of mental models that helped him think about the world. Compound interest, confirmation bias, & something he calls the Lollapalooza effect - human psychology that shapes human behavior. Tulips, cryptocurrency, & cola all fit into this effect.

The typical software startup raised their Series A 15 months after raising their seed at 2x their seed valuation.

A year ago, that Series A would have been raised three months earlier at 3.5x the valuation.

Here’s another way of visualizing the data. The red dot is 2023 - the other years are in grey. Across rounds for the 50th & 75th percentile of companies, step-up valuations are the lowest multiples in about ten years.