This is genuinely wild.
Cloudflare just dropped new Radar data showing that bots and AI traffic now account for 57.5% of all HTML webpage requests on their network.
Humans are down to 42.5%.
For context, Cloudflare handles around 20% of the entire internet. So this is not a small sample size.
Their CEO said the agentic AI wave has flipped the script years ahead of their 2027 forecast. Crawlers, scrapers, autonomous agents. Hammering sites non stop.
Most of the visitors reading content and loading pages right now are machines.
This breaks a lot of things at once.
Ad models that priced eyeballs, SEO playbooks built around human reading patterns, site design and rate limits built for human behavior. All of it needs to be rewritten for machines first.
The internet just stopped being a human-first place.
We are the minority now.
Meanwhile Trump's 🇺🇸🖤🇷🇺
"This will be one of the major infrastructure projects between our countries. As for the tunnel, we will have news tomorrow — we are signing an agreement to continue the design of the tunnel" - Dmitriev
@4n4lisis
Worst energy shock in history transpiring; inventories about to bottom; Germany buying BC LNG; South Korea, India and China all come to Canada in the same two week period asking for oil and gas; TSX O&G index up 35% YTD.
The Canadian Left:
🤡
SoftBank's €75B AI datacenter bet specifically in France is the market pricing what the AI power constraint chain actually looks like. France = 60%+ nuclear, lowest industrial power costs in Europe. EU industrial electricity runs 2x US prices — that delta is the moat. No SMRs operational outside China/Russia. Data center operators are locking in 10-year power deals now because they know the constraint chain: gas turbines and grid interconnect today, nuclear construction lead times tomorrow.
https://t.co/TfGHj4JadG
This is why there should be a corollary to Jevons Paradox: exponential growth attracts disruptive innovation.
When demand and costs increase in unison the incentives to “fix the problem” increase super-linearly (Δ demand x Δ cost). The prize becomes larger and attracts mental and financial capital.
This is what happened with fracking and led to the oil bubble burst in 2008. Higher unit prices meant previously uneconomic techniques were suddenly profitable. Moreover, the new techniques became more affordable over time – following the disruption framework that Clay Christensen lays out in The Innovator’s Dilemma (and that @eastdakota built Cloudflare on).
The best video you can watch on this is from @JonathanRoss321. Well worth the time!
The market priced software automation as code obsolescence. I’ll take the other side of that trade.
AI is transforming software from a human-driven activity into a machine-driven consumption engine operating 24/7.
$DDOG $NET $MDB $SNOW
The Market Is Wrong About AI Software and Bitcoin
The early take on AI was simple: if agents replace workers, and workers use software, then software demand falls. Clean logic. Completely wrong.
We have seen this mistake before. When the internet arrived, the consensus was that paper would disappear. Instead, paper usage rose for years because the internet made it easier to create, copy and distribute information. Demand expanded, it did not shrink.
AI is doing the same thing to software.
Agents do not replace software. They consume it, constantly. An AI agent is not just a chatbot. It reads information, makes decisions and calls tools to execute tasks. Databases, spreadsheets, CRMs, code engines, each step triggers another action, another request, another loop.
A human uses software during the workday. An agent uses it 24 hours a day, seven days a week, without pause. Software activity is already exploding, with a growing share driven by AI agents rather than people. What looks like automation is actually a surge in usage.
When usage rises, spending follows. Companies deploying agents are not shrinking their software budgets they are expanding them. The model is shifting from paying per seat to paying for consumption. The market, as usual, is misreading the shift.
Don’t even get me started on AI and the labor force.
Capital is crowding into a narrow set of semiconductor names, as if every company in the space shares the same economics and exposure. They do not. Memory is not logic. And not every Semi stock needs to be re-rated.
Cyclicality is not scarcity. Use that. The stupidity of the market is an opportunity. While FOMO traders chase anything tied to AI and assume every semiconductor stock looks like Micron, the real differentiation is in who actually captures sustained demand.
The same analytical error is now showing up in Bitcoin.
Whether the Clarity Act passes or not, or whether stablecoins offer yield or not, does not matter to Bitcoin. Those debates are distractions. They do not change its role, its design or its long-term demand.
In both cases, the mistake is identical: confusing what changes at the edges with what drives the system.
In AI software and in Bitcoin, the edge trade is noisy, the base trade is obvious, and that is where the real money is made.
Documenting the headwinds I now see for AI.
It won't seem like it, but I love AI and am long-term positive. But when "math doesn't math" I take note.
1. The core thesis for foundation model lab investment has been high upfront investment made worthwhile by significant long-term profits.
2. These are capital intensive businesses and the compute commitments are very high relative to revenue and require strong growth over long time periods. The "leverage" (commitments versus revenue) is extremely high.
3. The fundamentals are not as positive as they previously were:
• Input costs are higher (commodities, chips, power)
• Interest rates are higher
• Competition is more intense
• Scaling Laws are now problematic: exponential costs/power cannot continue
4. Forecasting compute spend is challenging and high risk due to (a) revenue uncertainty and (b) algorithm uncertainty
5. Revenue growth appears to be slowing. The technology is valuable, but ROI is proving to be more expensive and take longer than anticipated.
6. The future is likely "different models for different use cases" with the lower end of the market being highly competitive.
7. Core use cases such as agentic software engineering are likely to need approaches beyond next-token prediction. They are Σ₂ᴾ complexity problems requiring multi-objective optimization and likely a combination of Transformers and other methods.
8. Current forecasts in memory makers are built largely on quadratic attention. That will not persist: we are already seeing work from DeepSeek, Minimax and Nvidia that can cut RAM needs by 80% or more.
9. This means semiconductor valuations are substantially overinflated and will go through the traditional glut versus shortage cycle.
10. For foundation model providers: lower costs with competitive differentiation is good. However, lower costs with a lack of differentiation would mean lower revenues. This makes it harder to (a) service commitments and (b) pay back investors.
11. Leverage is substantially higher than in previous cycles, evidenced by leveraged ETFs, call option activity and margin loans. Korea is particularly susceptible.
12. 0DTE options create a profile that has stronger parallels to portfolio insurance and 1987 than any other point I can remember.
13. The combination of exponential increases in call activity coupled with the ties of semiconductors to structured products means there is a non-trivial systemic risk to the financial system.
14. Implied earnings growth rates are inconsistent with other periods in history.
15. Macroeconomically we cannot and should not fund exponential cost increases. History has shown us repeatedly that there are better ways (see Quick Sort and Simplex).
16. Significant supply is hitting the market via IPOs.
––
Taken together: costs and competition are increasing while revenue growth is likely slowing. Valuations are fragile and prone to technology disruptions that are already here. Systemic financial market risk is extremely high.
Oil predictions, over the next 10–20 years:
1- oil’s share of energy drifts toward 25%
2- absolute demand stays flat to slightly up
3- mix shifts further from fuel to materials and other hard-to-replace uses
Less demand elasticity, supply more political, and price more volatile
The most expensive misread in markets right here is thinking AI kills software. I think it gets absorbed by it which is a completely different trade.
Spent several weeks on this post and built a 15 name basket around the idea.
https://t.co/2jGQf1SK2J