New Ask Claude or ChatGPT about any stock, using Bullpen. Bullpen in Claude or ChatGPT. 7 days free

Wednesday, September 30, 2026

Back to the front page

The Next Consumer is a Computer

Meta’s Muse illustrates how persistent AI agents shift compute from per-query to per-user, driving heavier GPU, CPU and memory demand.

From X

Convequity on X

$PANW and $DDOG both put something real inside the customer's systems that an AI lab cannot recreate by calling SaaS APIs. Palo Alto's install is still harder to replace with AI. 1. Datadog's agents mostly sit outside the traffic and watch production. A bad agent version rarely takes the business down the way a bad firewall rule can. 2. Palo Alto's firewalls and SASE cloud gateways sit inside the traffic and decide what to allow, deny, decrypt or inspect. A wrong allow or deny hits live traffic and live machines. 3. Many Datadog users start the day in a coding agent and open Datadog when something looks wrong. Many Palo Alto users start in its console and stay there through alerts, triage and response. For them, an outside AI agent is an add-on. Cybersecurity work splits two ways on AI risk. Work done before code ships ("shift-left") or from outside the traffic, such as scanning code, checking cloud settings and sorting collected alerts, gives answers that are easy to check. AI agents are already taking over more of it. Work in live traffic and on live machines ("shift-right") means judging whether an attacker is still inside and whether blocking them is safe. That is more ambiguous and mistakes cost more, so enterprises move slower and demand earned trust. Palo Alto is not risk-free. In our view its exposure is the easily checked detection and vulnerability work inside its platform, which AI agents can take over sooner. The threat is not an AI-native firewall startup.

Convequity on X

Anthropic's biggest near-term threat isn't the S-1 or AI-safety language. It's competition, especially an OpenAI price war. The S-1 is public, and the timing lines up with OpenAI DevDay. My read is that it isn't a coincidence. That's my read, not something either company has said. What surprised me more came out of DevDay: OpenAI is ending its price war. Here's what that means in practice. OpenAI removed the Pro 20x subscription at $200/month and plans to cut usable tokens on that $200 plan by about half. The allowance was also more generous than it looked. Over the past quarter, OpenAI did roughly 2.5 manual usage resets per week. So heavy users didn't only get a big allowance. It got refilled before they hit the wall, over and over, for about three months. That's a price war fought in tokens instead of list prices, and it is expensive to fund. Why it mattered for Anthropic: for heavy users, cheaper tokens beat a better model. Once a model is good enough, what limits you is how much work you can push through it per dollar. A better model you have to ration loses to a good-enough model you can run all day. Anthropic went the other way and raised margin. That's the right posture for an S-1. It's the wrong posture when a competitor is paying for volume. That left Anthropic in a trap going into the IPO: - Match the pricing, and margin thins just when margin is the story you're selling. - Hold the line, and growth slows as heavy users drift to wherever tokens are cheapest. Either way the numbers get worse at the worst possible time. Ending the price war takes that trap off the table for now. Anthropic gets a cleaner tape into the IPO. Why did OpenAI stop after one quarter? I see two reads, and both are mine, not confirmed: 1. The burn. My read is that OpenAI postponed its own IPO because of how much it was burning, and investors pushed it to stop. 2. OpenAI's own later IPO. If it wants to list, its numbers have to look like a business too. One quarter of subsidized tokens is a campaign. A year of it is a problem in your own filing. The second-order point is that in both reads OpenAI stopped because of its own finances. Anthropic didn't win the war. It ended because the side paying for it chose to stop. The playbook still exists and it ran for a full quarter. It can be turned back on. That's the risk that remains. OpenAI restarts it. Or a rival like $SPCX xAI ships a next-gen model and discounts tokens hard to buy share. It's a bit like Uber's cash burn: a well-funded player can keep subsidizing longer than a margin-focused company wants to wait. I don't think that's fully priced in.

Microsoft: The Xbox Segment Revamp

Microsoft plans to rebuild Xbox after FY2026 through studio cuts, higher hardware prices, Game Pass growth, exclusive releases and the 2028…

Toll roads in the sky

Asia's elevator boom favors maintenance and modernization providers, with Otis deriving nearly 90% of operating profit from after-sales…

EUXTF · long

Euronext NV

Euronext has evolved from a regional cash-equities exchange into a pan-European market-infrastructure platform after a decade of technology…

CoreWeave, Inc (CRWV)

CoreWeave funds GPU data-center expansion with debt backed by $104 billion in contracted revenue, while GPU cloud rentals generate roughly…

GameStop Is Extreme Deep Value

GameStop's collectibles mix rose to 45% of sales as gross margin reached 44%, supporting a deep-value case built on cash and store…

Publications we follow

Every publication on this site, hand-picked.

130 publications

Newsletters 110

X accounts we follow

Every account on this site, hand-picked.

91 accounts

Investor letters we follow

376 firms

Sectors

Ordered by how often our editor reaches for them.