# The 10 Biggest Days in AI History
*The AI Daily Brief — Thursday, 2026-11-19 · https://aidailybrief.ai/e/2026-11-19*

**Anthropic's IPO ends AI's private era and starts its public one.**

Generative AI's biggest days so far have mostly happened inside private companies. A nonprofit founded out of fear of Google, a transformer paper Google underused, a chatbot Anthropic sat on, a boardroom coup, a model that made agents real. Along the way the public kept pushing in: through Nvidia's crash on DeepSeek, through Washington's fights with Anthropic, and through a midterm vote against data centers. With Anthropic now public, ordinary people can buy into AI's builders directly and those builders face a new degree of transparency. What happens on day one matters much less than what comes next.

---

## By the numbers
- **5 days** — Time it took ChatGPT to reach 1 million users
- **100M** — ChatGPT users within five weeks, the fastest-growing app ever
- **$589B** — Nvidia's one-day DeepSeek loss, the biggest single-day loss for any company
- **17%** — Share of Nvidia's market cap wiped out on January 27, 2025
- **~800** — OpenAI employees who nearly all signed the letter demanding Altman's return
- **18 days** — How long Anthropic's model stayed dark after the government's June directive
- **56%** — Voters who called data centers a major or crucial factor in their midterm vote (NBC exit poll)
- **45 of 46** — Local ballot measures on data centers that the pro side lost
- **$83** — Anthropic's IPO price, above its $70–78 range

## Main episode

### Anthropic IPO day is the start of a very new era `[01:00]`
Anthropic's first day of trading makes it officially the biggest IPO in history. NLW calls it the culmination of many AI storylines, which is why it's the right moment to look back at the other days that got us here.
*For: Finance, Exec*
Link: https://aidailybrief.ai/e/2026-11-19#anthropic-ipo-day-new-era

### OpenAI was founded out of fear of Google `[02:00]`
OpenAI launched as a nonprofit on December 11, 2015. Sam Altman's pitch to Elon Musk was that AI would be built anyway, so it shouldn't be Google that got there first. Musk had already put about $5 million into DeepMind partly to keep an eye on Demis Hassabis. Backers committed a headline $1 billion to advancing digital intelligence 'unconstrained by a need to generate financial return.'
Link: https://aidailybrief.ai/e/2026-11-19#openai-founded-to-beat-google

### The nonprofit constraint broke almost immediately `[03:00]`
NLW pairs the founding with June 22, 2019, when OpenAI announced a $1 billion investment from Microsoft. OpenAI had concluded it couldn't raise enough as a nonprofit and created a for-profit arm. That move set off many of the fights that came later.
*For: Finance*
Link: https://aidailybrief.ai/e/2026-11-19#microsoft-billion-for-profit-arm

### Google invented the transformer, and OpenAI saw what it was `[04:00]`
On June 12, 2017, Google Brain published 'Attention Is All You Need,' introducing the architecture behind modern LLMs. Google mostly applied it to search queries and Translate, while OpenAI quickly recognized it as what they'd been looking for. Co-author Noam Shazeer later left Google in 2021 to found Character.ai after Google refused to release his transformer chatbot, Meena.
*For: Eng*
Link: https://aidailybrief.ai/e/2026-11-19#attention-is-all-you-need

### Anthropic had a chatbot first, and a rumor about it triggered ChatGPT `[05:00]`
In February 2022, Anthropic was already testing an internal chatbot called Claude but held it back for fear of starting an arms race. That November, OpenAI was preparing GPT-4 when a rumor spread that Anthropic was about to ship a chatbot. OpenAI rushed out its own, announcing it to the team on November 17 and shipping on November 30.
*For: Product*
Link: https://aidailybrief.ai/e/2026-11-19#anthropic-sat-on-claude

### The team expected tens of thousands of users; they got a million in five days `[06:00]`
Most of OpenAI expected ChatGPT to reach the low tens of thousands of users, and Sam Altman's most optimistic guess was an eventual million. It hit a million in five days and 100 million in five weeks, becoming the fastest-growing application of all time. ChatGPT quickly became synonymous with AI.
*For: Product, Marketing*
Link: https://aidailybrief.ai/e/2026-11-19#chatgpt-growth

### The four-day coup that made OpenAI look beatable `[06:00]`
On November 17, 2023, OpenAI's board fired Sam Altman for not being 'consistently candid,' and never clearly explained what that meant. Interim CEOs changed twice and Microsoft offered to rebuild the team in-house. Almost all of OpenAI's roughly 800 employees threatened to leave, and by November 29 Altman was officially back with a new board.
*For: Exec*
Link: https://aidailybrief.ai/e/2026-11-19#altman-ouster

### The real legacy of the Altman saga: Microsoft started hedging `[08:00]`
NLW argues the drama changed the industry's structure. OpenAI no longer looked unassailable, and the returning Altman was never going to be as powerful as before. Microsoft's access to OpenAI models ended whenever the board declared AGI, which looked much scarier after this episode. That set in motion the Mustafa Suleyman hire and Microsoft's slow move away from OpenAI.
*For: Exec*
Link: https://aidailybrief.ai/e/2026-11-19#coup-reshaped-microsoft

### Claude 3.5 Sonnet was the day Anthropic stepped out of OpenAI's shadow `[12:00]`
On June 20, 2024, Claude 3.5 Sonnet broke new ground in agentic coding and let Cursor, Replit and Lovable bring that capability to non-coders. From then on, Claude models basically led on coding. Coding turned out to be the foundational use case that many other use cases were built on.
*For: Eng, Product*
Link: https://aidailybrief.ai/e/2026-11-19#sonnet-35-coding-lead

### DeepSeek erased $589 billion of Nvidia in one day `[13:00]`
On January 27, 2025, fears about a Chinese lab sent Nvidia down 17%, the biggest one-day loss for any company ever. The panic rested on a viral claim that R1 was trained for $5.5 million, a figure that turned out to be wildly overstated. DeepSeek's free app was also number one on the App Store.
*For: Finance*
Link: https://aidailybrief.ai/e/2026-11-19#deepseek-nvidia-crash

### Most people's first reasoning model came from China `[14:00]`
R1 wasn't the best model in the world; it was a reasoning model like OpenAI's o1. Almost no one had used o1, which spent months in preview and then sat behind paid plans. Nvidia recovered, but the DeepSeek moment changed how Western labs released models and started the US-versus-China framing that has shaped policy ever since.
Link: https://aidailybrief.ai/e/2026-11-19#deepseek-reasoning-gap

### Opus 4.5 made the agentic dream real `[14:00]`
Agents were all over the discourse in 2024 and 2025 but not really real until Anthropic released Claude Opus 4.5 on November 24, 2025. It took the community about a month, and some holiday downtime, to realize how different it was. NLW credits the pairing of Opus 4.5 with Claude Code as the true accelerant.
*For: Eng*
Link: https://aidailybrief.ai/e/2026-11-19#opus-45-agentic-era

### From Clawdbot to Molt to OpenClaw in a single frenetic week `[16:00]`
Peter Steinberger's open-source agent Clawdbot felt like the first agent that could actually do things for you. Anthropic sent a cease and desist over the name, so it became Molt, and an entrepreneur launched Moltbook, a social network where agents pondered consciousness, griped about users and started a religion. On January 30 the project settled on its final name, OpenClaw, and the agent craze went global.
*For: Product*
Link: https://aidailybrief.ai/e/2026-11-19#openclaw-naming-saga

### Genuinely the most incredible sci-fi takeoff adjacent thing I have recently seen. `[17:00]`
*— Andrej Karpathy, on Moltbook*
Many Moltbook posts may have been humans posing as agents, or agents role-playing consciousness. It still caught enormous attention as a vision of what the future might look like.
Link: https://aidailybrief.ai/e/2026-11-19#karpathy-moltbook

### Agents broke Wall Street's bubble math `[17:00]`
In fall 2025, the AI-bubble fear came down to arithmetic: $20 or $30 seats times the number of available workers didn't add up to the giant infrastructure deals. Once individual agentic users were spending hundreds or thousands of dollars a month, the industry's whole financial picture changed. NLW's view is that Claude Code, Opus 4.5 and OpenClaw together split AI into a pre-agentic and an agentic world.
*For: Finance, Exec*
Link: https://aidailybrief.ai/e/2026-11-19#agentic-economics-bubble

### The first US company ever labeled a supply chain risk `[18:00]`
On February 27, President Trump ordered federal agencies to stop using Anthropic, and Defense Secretary Hegseth barred it from military use. The trigger was Anthropic's pushback over its models being used in the January raid on Maduro. Its red lines were no autonomous weapons control and no domestic surveillance. Anthropic is still fighting the designation in court.
*For: Legal, Exec*
Link: https://aidailybrief.ai/e/2026-11-19#anthropic-supply-chain-risk

### The end of labs shipping models whenever they choose `[19:00]`
Anthropic gave its powerful Mythos model a private period under Project Glasswing, letting partners use it to harden systems before public release. After the public version launched in June, Amazon warned the government about a jailbreak. A June 12 directive ordered access cut for foreign nationals, and because Anthropic couldn't build identity checks in time, it pulled the model for everyone for 18 days.
*For: Legal, Eng*
Link: https://aidailybrief.ai/e/2026-11-19#government-pulls-model

### The midterms became a referendum on data centers `[20:00]`
Backlash built all year over noise, tax breaks, electricity prices, water use and NDAs, and a May Gallup poll found more opposition to a local data center than to a nuclear plant. On November 3, 56% of voters told NBC that data centers were a major or crucial factor in their vote. The pro side lost 45 of 46 local measures, and Democrats flipped both chambers.
*For: Legal, Ops*
Link: https://aidailybrief.ai/e/2026-11-19#midterms-data-center-referendum

### Not a ban hammer, but a demand for a voice `[21:00]`
NLW expects stricter standards and processes rather than a shutdown, and still thinks data centers will get built. He calls it an extremely winnable issue for everyone. The bigger signal, he argues, is that people showed they won't let the future of AI be written without them.
*For: Exec*
Link: https://aidailybrief.ai/e/2026-11-19#data-centers-winnable

### The real significance: AI goes from private to public `[22:00]`
Anthropic priced on November 18 at $83, above its $70–78 range. People can now invest in AI's builders directly rather than through proxies like semiconductors, and being public forces more transparency. One bad quarter of AI demand could ripple through an economy this tied to AI, but NLW thinks being public is one of the best ways for outsiders to gain influence over these companies.
*For: Finance, Exec*
Link: https://aidailybrief.ai/e/2026-11-19#private-to-public-era

### ChatGPT might not have gone as far without Stable Diffusion `[23:00]`
The first honorable mention is August 22, 2022, when Stable Diffusion went live. AI-generated images flooding social media helped fuel the late-2022 AI moment alongside ChatGPT.
*For: Marketing*
Link: https://aidailybrief.ai/e/2026-11-19#honorable-stable-diffusion

### o1-preview answered the plateau question `[24:00]`
For much of 2024, the worry was that training methods had hit a wall, which had big implications for AI's disruption and business model. OpenAI's o1-preview on September 12, 2024 showed that reasoning models could squeeze much more out of existing approaches without a new pre-training breakthrough.
*For: Eng*
Link: https://aidailybrief.ai/e/2026-11-19#honorable-o1-preview

### Two dates that sum up the AI trade and the build-out `[24:00]`
Nvidia's May 24, 2023 earnings call began a pattern of the company redefining quarterly performance that continues today. Project Stargate's White House announcement on January 21, 2025, with Altman, Ellison and Son, got the infrastructure story rolling, even though bigger deals have followed.
*For: Finance*
Link: https://aidailybrief.ai/e/2026-11-19#honorable-infrastructure

### Jensen Huang's resignation is the top contender to join this list `[25:00]`
Huang resigned as Nvidia CEO on November 12, 2026. It could reshape AI financing given Nvidia's role as the industry's central bank, along with AI policy advocacy and US-China relations. It could also turn out to be a nothing burger, but NLW calls it the likeliest date to climb into the top 10 in a year or two.
*For: Exec, Finance*
Link: https://aidailybrief.ai/e/2026-11-19#jensen-resignation-wildcard

*Today's sponsors: KPMG, Granola, Gartner, Blitzy — offers at https://aidailybrief.ai/sponsors*

---
Transcript: https://aidailybrief.ai/e/2026-11-19/transcript.md
Listen: https://pod.link/1680633614 · Ad-free: https://patreon.com/aidailybrief
© 2026 The AI Daily Brief — Until next time, peace ✌