Apologies, this issue is a day late. I've spent extra time researching and having conversations with other founders to ground my assumptions. I hope today to provide useful information that does 2 things:
- Help people prepare
- Highlight profitable opportunities With that said, let's get into it.
I've spent over a year working with AI every day. I have spent countless hours researching the impact it will have on the markets and economy as a whole. I have built countless products and automations and I've seen the value and limitations of AI.
I don't have all the answers, but there are important signals I'm seeing that I intend to highlight in this week's issue.
Is AI in a Bubble?
Yes. I'll ground this answer shortly, but we are in an AI bubble and it's important for people to prepare for it to pop. It's already starting, but it's not what you might think it is.
AI Is Here to Stay
This is important to acknowledge before we dive into the details of the AI Bubble. Dotcom was a bubble, but domain names are still around today. So first and foremost it's critical to understand that AI isn't going anywhere. It is here to stay and you should learn how it works. There was a "golden era" for web developers for a while. Now AI builds website exceptionally well, but there will be another golden era for those who master this technology.
What is the Bubble?
If you've been working with AI and even building products, you will have seen how difficult it is to generate consistent and profitable revenue. Most of the revenue in AI right now is going to the large AI providers to rent their models. Many AI products and implementations do not turn a profit, or the company can't measure a noticeable impact.
MIT's NANDA initiative found that 95% of enterprise AI pilot programs delivered zero measurable P&L impact, despite $35-40B in collective investment. A separate NBER study of 6,000 executives across four countries confirmed that nearly 90% of firms report no productivity gains from AI so far.
While there is a significant amount of revenue in "AI" most of it is captured by Anthropic and OpenAI, who are both burning billions and have yet to turn a profit.
What's crazier is not OpenAI and Anthropic's burn rates, but the amount of investment capital being poured into this. VCs are AI Hungry, investors are pouring billions into AI businesses, products, model development, and AI data centers, and I've never seen so much money going to pre-seed founders with an AI idea.
In Q1 2026 alone, $242B went to AI companies. That's 80% of all global venture capital for the quarter. Four companies (OpenAI, Anthropic, xAI, Waymo) raised $188B between them. For the full year of 2025, AI captured $211B in venture funding, up 85% from 2024. The OECD confirmed that AI now represents 61% of all global VC investment.
Model training and model serving have made power one of the top global line items for largest expenses. The data centers are expensive to build and consume enormous amounts of energy.
This is the bubble. And it's starting to pop.
What The "pop" Looks Like
We are in a bubble of investor subsidies, cheap AI, and failed pilots. What does the pop look like?
This is where it differs from the dotcom bubble. That crash was hard and abrupt for many companies. This bubble will still hurt, but it's going to look different.
Investors want their money back and many over-extended. There will be billions in lost investments in the US alone for failed AI startups and products. I don't believe OpenAI will last, which will result in over $120B in investment capital loss.
Google just partnered with Anthropic, allowing them to use their TPUs which are a fraction of the cost of NVIDIA GPUs. OpenAI is going to be paying 5-10x more on hardware than Anthropic, and Microsoft has started developing their own frontier models. That's a strong signal that OpenAI is operating on a short runway.
These losses are significant and will be felt throughout the entire economy as a whole. But more than that, Anthropic is already starting to reduce their limits and increase their pricing for Claude. It's time to turn a profit.
AI has been heavily subsidized by investor cash and that's coming to an end. OpenAI burns $1.69 for every dollar it earns. Anthropic is already reducing usage limits and raising prices. To become sustainable, these companies will need to at minimum double their pricing, and that's a conservative estimate. This is a sting that will impact every business using AI.
The pop is not going to be some dramatic explosion and death to AI. It will be a sharp increase of AI pricing that will drastically limit overall use cases.
Preparation
One of the reasons Ryker and I are building Courier is to prepare for this. Apple and Google have bet on infrastructure over models. The M-series Macs are the most power efficient and affordable compute option for hosting LLMs. We are building solutions that allow people to use AI just like they would with Anthropic or OpenAI but on their Mac instead of renting from these companies, protecting themselves from cost volatility.
If you aren't self-hosting or haven't experimented with it yet, I encourage you to create a Courier account and start using our free plan to familiarize yourself with Open Source models. This is one of the best things you can do to protect yourself, but compute costs will rise as more and more people turn to his option to combat rising AI costs. Apple is already experiencing slight price increases.
Opportunity
THE FUTURE OF AI ISN'T BETTER MODELS, IT'S BETTER SYSTEMS.
I can't say this loud enough. This is where the biggest opportunity is. Models are getting better and smaller every month. Pretty soon people will be using models exclusively on a 64GB MacBook Pro.
The problem is most solutions are built leveraging the biggest, most capable models available. This is incredibly expensive to use and expensive for these frontier labs to chase.
Golden Rule of AI
The golden rule is this: systems + structure > bigger models
Good systems and the structure offload cognitive load, resulting in better, more consistent outputs with lower intelligence requirement. Small models are incredibly capable with the right scaffolding.
The AI solutions that are winning right now are doing 3 things:
- They provide measurable value — (saved time, revenue, cost optimization)
- They are affordable and predictable — (a SMB is much more likely to subscribe for $100/mo instead of usage based pricing)
- They are using structure and systems to get away with smaller or self-hosted models that reduce or eliminate the token tax
If you are looking at running an AI automation agency or building an AI product this is where I recommend you spend all of your time and energy. Answer these 3 questions well and you will be able to sell a strong solution. Communicate this to prospects and you'll get customers. Protect your margins and your customers from the pop.
Wave
The white-collar job economy is in a strange place right now. In Q1 2026, nearly 80,000 tech workers were laid off with almost half explicitly attributed to AI and automation. In 2025, companies cited AI in 55,000 job cuts, a 12x increase from just two years earlier. Here's the problem: HBR published a piece in January titled "Companies Are Laying Off Workers Because of AI's Potential — Not Its Performance." The NBER data backs this up. 90% of firms saw no productivity gains. Companies are cutting people based on a bet, not a result.
I've never seen a wave of such talent left so jaded and untrusting. I see a revolution happening. A revolution from big tech and soulless corporate culture.
Never in my life have I seen so many solo agencies and solopreneurs, with so many being quite successful. I experienced this first hand working at a company with a terrible corporate culture that started creating false complaints about employees to dodge their own benefits policies.
There are now 29.8 million solopreneurs in the US, contributing $1.7 trillion to the economy. Solo founder rates on Carta have doubled over the past decade, hitting 36% of all new startups in 2025. A record 5.6 million independent workers earned over $100,000 last year. LinkedIn reported a 69% jump in people adding "founder" to their profiles, with 47% saying AI makes them more likely to start a business. After the 2023-2024 layoffs, 69% of employers hired freelancers to fill the gaps those full-time employees left behind.
I think we will see an interesting wave of contractors and solopreneurs that disrupts traditional big tech in a major way. And AI makes this more possible than ever. I say that in the context of the bubble and the prep work necessary to shield from increasing costs.