Jeff’s Predictions for 2024

With the end of 2023 and the start of a new year I’m ready to peer into my magic 8-ball and give my predictions as to what 2024 may hold…

Bullshit tech jobs continue to go away

Two notable things happened in tech in 2023 that resulted in mass layoffs this year:

1. Cost of capital/interest rates remained high following a run up of rates in 2022.
2. Elon fired 80% of Twitter, and the app seems to still work just fine*.

Love Elon or hate Elon, he was the first CEO of a large tech company that laid off a large amount of the workforce. While its debatable as to if this move served contagion for other tech companies to follow suit, its undeniable 2023 was rife with tech layoffs. As of December 2023, Inc reports 1,169 tech companies had let go of 260,238 throughout this year. This is a trend I believe will remain sticky for three big reasons.

First, anecdotally, the best work I’ve done professionally has been when I was a part of a very lean & very high output group of three to seven individuals. This is one reason I’ve made the decision to dedicate this phase of my career to building cool things at bootstrapped or Seed/Series A/Series B level companies. Though it may seem counter intuitive, adding more people to a team often slows velocity and gums up the “product machine”. A handful of “9” or “10” quality employees are multiple orders of magnitude more effective at making world changing gains than 50+ “5” or “6” quality employees (or 50+ of those employees surrounding a handful of 10s). To add, the whole purpose of technology is doing a more with less. You shouldn’t need big teams to do much of anything around pure play software.

Second, we are learning just how impactful certain AI copilots can be for more technical roles. I anticipate this trend to continue, making it all the more obvious that lean teams doing great work is always going to be better for the company’s bottom and top line than adding a bazillion people to middle management positions. When money was seemingly free, the moats large and we were all drunk on 0% rates it felt okay to hire like a drunken pirate. Those days are well in the rear-views.

Third, we’re learning how bullshit a number of positions at technology companies really are. TikTok used to be full of “day in the life” videos from a mid level PM at Meta, Twitter, or Google that looked more akin to a middle class vacation than a value-add job. Tech companies are wising up to this and going back to old school hard nosed principals. I’ve long subscribed to the Pareto Principal, and it seems this is something that is on steroids at many of your best software companies. 20% of the employees do 80% the work….except its more like 5-10% do 90%-95% the work. Its usually not hard to figure out who the rockstars are and who the dead weight is, which is why I’d expect a lot of this headcount reduction to remain sticky (or even continue perpetuating).

Translating this into real world terms, its probably time to ask yourself “what is it you’d say, you do here?”; if you can’t draw a direct line between your efforts at a company and legitimate value being added (moving stuff around on a Jira board does not count as “value add”), I’d start looking to pivot, start adding skills to your educational backlog or consider something all together new.

_{*Twitter is very much not profitable, but the tech stack didn’t crash and the app still works despite massive headcount reductions.}

The Outdoor Industry’s Lag Continues

Twenty 23 was a weird year for the outdoor industry. The COVID supply/demand mismatch created a whipsaw effect that left far too much inventory around through the latter half of this year which has led to pricing plummeting, especially in sports like mountain biking. Add to this rising interest rates, an increase in the cost of capital and a leveling of product innovation and you’ve got all the makings for a massive slowdown in the broader outdoor/action sports industry.

Do not misread, the industry is not in “trouble”. However, companies like Guerilla Gravity, VanMoof or even REEB all have had to (or will) take some very sobering medicine when it comes to recognizing that growth at the expense of profitability does not work in 2023, and will quickly lead to your dismantling. If you cannot be profitable now you either need to cut burn or (if that doesn’t work) close shop. You won’t be raising capital, barring a really rich uncle or buddy who sold his tech company in 2020 being affiliated with the firm.

Innovation is another lesser talked about part of this equation. Gone (for now) are the days in skiing, mountain biking, snowmobiling, dirt biking or ______ (name your sport)____ where buying the latest greatest really gives you an edge. The pace of innovation has drastically slowed, this was evident in skiing in the early 2010s and mountain biking/snowmobiling right around 2020. While the gear head part of me might frown a bit, the sportsmen in me loves this because it allows for an equalization based on skill, not size of bank account.

Jeff’s Anecdotal Examples ‘Proving’ This

1. Mountain Biking: I bought the same exact enduro bike twice, and would buy it a third time (Specialized Enduro) if I was racing very seriously. We are chasing far smaller positive iterations in the sport of mountain biking compared to just a handful of years prior.
2. Dirt Biking: My hot new 300XC is a great bike, but as proven by Jarvis (who still rides the TPI), plenty of people like the prior gen 300 just as much. The old frame was less rigid, there are no stator issues, power delivery smooth and flywheel in all models more appropriate for harder stuff. Simply, the the old one was a hell of a bike and as much praise as the new one has received, it doesn’t change your experience on trail..
3. Snowmobiling: I’m in no hurry to upgrade my Patriot Boost and see nothing on the horizon that is significantly better in any way. In fact, I don’t even think the aftermarket stuff is worth considering for the snowmobile. Its just flat that good…and now going into its 3rd season.
4. Skiing: I routinely buy skis that are 3-4+ years old. In fact, some of my all time favorite skis are 10-15+ years old (this is a link to the Praxis Powder board – my favorite pow ski and is from 2006) Pace of innovation was incredibly rapid in he late 2000s and early 2010s but now is all but gone.

Startup Fundraising Goes Lean & Venture Capital Hits a Slowdown

Software has eaten the world. While I know there are a myriad of problems to still solve with technology, I strongly question if the venture model of old will continue to hold as much water into the 2020s. Smaller teams bootstrapping profitable companies aiming at smaller markets, as opposed to giant companies raising hundreds of millions aiming at massive markets operating at a loss, will become much more popular as we aim down the barrel of this decade.

To best articulate this, its first important we briefly talk about how VC style investing & fundraising works. In a nutshell, a venture capital firm invests in a portfolio of startups expecting most to fail. However, the minority that are successful are expected to more than makeup for the losses the VC firm takes and generate the strong risk adjusted returns the top tier firms are known for. I’ll use an example from Silicon Valley lore here to articulate how this work…

When eBay had just found product market fit, Benchmark Capital invested $6.7M into the auction site company for a ~22% stake (ironically eBay never spent this capital). This $6.7M investment became worth around ~$5B at its peak or 700x the original investment amount. Hence, Benchmark’s original fund of $85M was highly successful on this one investment regardless how the other investments in the portfolio did. This is very much how this style of investing works; one or two investments in an entire fund can literally make for its success. Hence you need the possibility of extraordinarily outsized returns against the amount you invest for this model to work. If founders begin aiming at smaller markets, smaller exits and require less capital to get there, this is bad for the VC ecosystem. I predict we’ll see fewer Snowflakes, Ubers, and Airbnbs and instead see more “smaller” versions of MailChimp (which famously took on zero outside funding).

Remember founders, taking on capital is expensive if you are successful. Dilution sucks. Owning your company is part of the reason you started a company. You should always do the most with the least amount of capital required.

My big long winded point here is we’re going to see more ~10 person companies aiming at small markets, near immediate profitability and friends/family being the only capital requirements. Here is why.

1. The developer stack options & frameworks are well established: I don’t care what language or framework you want to use, there are so many great well supported options here from an engineering standpoint. This drastically reduces the time, energy and effort your engineering team has to put into building whatever software you are looking to build.
2. Infrastructure is agile and dynamic: This really goes with the above point, but between AWS, Azure, all the cloud DB providers and the ease in which an API call can get made, there are seemingly infinite tools at our disposal to build whatever we want at scale.
3. Small teams can now aim at precision niches: While previously you may have needed multiple teams with top tier devs to put together a fairly simple web+mobile app, you can now do this with a very lean high output tam. Add the AI copilots and we can anticipate a lot of bootstrapped companies aiming at very precise problems with smaller TAMs that otherwise would not have been plausible to go solve.
4. Profitability and being default alive is cool.
5. Interest rates are likely not going back to zero: Gone are the net zero “free” cost of capital days. This changes the multiples we might expect in software and the amount of risk an LP might be willing to take for a given return. In theory, this alone should make a certain percentage of companies be “unfundable”. (risk adjusted returns have to go up meaning you need companies to either be less risky or more likely to have a higher FV to invest in)

As someone who desperately wants to build a great company, I love the idea of being profitable fast, keeping my expenses in check, running the company as a business (IE, old school fundamentals matter) and am totally okay with a smaller $10-100M style exit (if you even want an exit…if the business is throwing awesome FCF…who cares?)

Editor’s Note: This isn’t to say VC opportunity won’t exist, it absolutely will, simply less of it. The 2020s will be punctuated by fewer large tech companies being built, lower capital requirements to build technology companies and more precise aim on behalf of founders. Survival of the fittest/Thunderdome time for the VC community.

Manual ICE Cars Become More Sought After

As much as I love my Model Y, I feel strongly we’re going to see car enthusiasts continue to drive (pun intended) toward manual cars of the 90s, 00s and 10s in mass as many recognize these are going to be in short supply in the foreseeable future. (from what I can tell there are less than 10 sports cars that aren’t exotics you can buy with a manual tranny in 2023) Whether these cars are economical to own or faster than their EV counterpart isn’t the point. A manual gearbox with 5 or 6 speed transmission and clutch is a throwback to yesteryear and a reminder to all of us just how enjoyable it can be to drive a twisty road or windy mountain pass. I’d argue we even see a number of these vehicles that we otherwise wouldn’t consider to be collectors items to rise in value (E36 M3, 04 WRX, etc)

LLMs will show (some) limitations

2023 was obviously the year AI went mainstream. One seemingly quaint Google paper led to one of the largest paradigm shifts we’ve ever seen in technology. To be fair, ML/AI was making great strides prior to the transformer model revolution, but the world will likely remember the introduction of the large language model in the form of the generative pre-trained transformer model (OpenAI’s GPT) as being the line in the sand where AI became “real”.

I’ve spent a lot of time thinking about what makes these new models so incredible, when I’ve seemingly been around other forms of AI my whole life. IE, I grew up with a calculator in my pocket and a chess engine that could reliably beat me. These are both “artificial” and “intelligent”, but they are very narrow in the intelligence they could provide. I can’t have a calculator spell anything except “boobs” and Chessmaster 3000 won’t help me understand quantum computing.

What makes GPT4 or any of the modern LLMs so amazing is their depth and breadth of knowledge. They continue to do things that they weren’t explicitly programmed to do. As someone who has long been fascinated by CS, this phenomena is a bit groundbreaking. While machine learning may provide insights we wouldn’t otherwise get to, the aim still needs to be tight for ML to work. This is not entirely the case with these LLMs, as we’ve seen a number of times such as when Microsoft researchers were able to get GPT 4 to draw a unicorn using obscure code despite not being programmed to draw or know what a unicorn was. How does it do this? Turns out, we aren’t entirely sure…

Our inability to fully understand how the transformer models work is what is so fascinating, scary and difficult all at the same time (MIT just put out a great piece on this here). We have built technology that is almost alien in nature and is forcing us to reconsider the methods by which we study these models. As a result, I predict we’re going to see a number of hard to solve problems arise as founders attempt to deploy fine tuned variations of these models into highly valuable verticals. Not only is the ground moving beneath our feat in a way that was akin to cloud computing in the mid 200s but the models also have difficult to solve problems including hallucinations or starting to eat their tail as they Frankensteining a homologation of prior genAI content into new content. Some of these problems can be solved with old school CS solutions, others are going to require all new and novel forms of thinking to effectively allow founders to deploy the “knowledge layer” within their stack.

As a result, I’d wager this will keep a lot of LLMs out of “mission critical” applications like medicine, certain types of finance/banking, military and certain transportation applications, which is unfortunately where they also hold the most promise. Don’t misinterpret here, anticipate a number of companies aiming at applying this technology to these industries (and would love to do so myself), but everyone will be doing so at a higher undertaking of liability the same way an aerospace company would take on additional liability using an engine that was from an alien species we didn’t fully understand. It may be significantly better, but there are substantial risks from this undertaking. One thing is for certain, anytime it feels like the Wild West, there is usually a lot of money to be made (and a lot of failure to also go with it…)

Lighting Round: Here are a handful of other predictions I don’t care to write about but wanted to put down…

1. Trump wins the election. (Editor’s Note; I’ve received a lot of feedback on this one. Please do not misread, I’m not suggesting this is a good thing, I’m just making a prediction. If it does turn out to be Trump v Biden 2024 – we all lose)
2. CRE may become a real contagion in the broader economy, especially for smaller regional banks who hold the debt
3. Rates may get cut more than we expect in part because of the national debt spiral we are in, inflation seems under control, Biden may put pressure on the Fed for reelection reasons and God Knows what other negative thing might happen that’ll lead to Daddy Powell pulling the rate reduction trigger.
4. Oil & gas production remains high in the United States; there is no substitute for highly energy dense easy to transport hydrocarbons for a number of purposes despite what you may read. I’ve already written about this, but I do not foresee oil and gas production waning.
5. You are better off being an electrician than an accountant. Blue collar style work and company building continues to be a great way to make money (with less of a threat from generative AI)