Home » Tech News November 19, 2026: Anthropic’s Record IPO, Gemini 4 Neon, and AI That Designs Chips
Posted in

Tech News November 19, 2026: Anthropic’s Record IPO, Gemini 4 Neon, and AI That Designs Chips

Glowing stock market chart rising over a city skyline with an AI neural network floating above it

Tech News November 19, 2026: Anthropic's Record IPO, Gemini 4 Neon, and AI That Designs Chips

It is a big day for AI — literally the biggest in stock market history. Anthropic starts trading on the Nasdaq today after pricing the largest IPO ever, Google just shipped another Gemini 4 model, Microsoft is trying to make AI cheap, Nvidia is walking away from China chip sales, and OpenAI's models are now redesigning the chips they run on. Here is everything that mattered in the last 24 hours.

Key Takeaways

  • Anthropic's IPO priced at $83 a share, raising $81.3 billion — the largest listing on record — and starts trading today under the ticker ANT.
  • Google's Gemini 4 Neon is a mid-tier model built for finishing long coding and knowledge-work tasks, priced at $3 per million input tokens.
  • Microsoft launched seven in-house MAI models at Ignite and an always-on Copilot Autopilot agent, betting that cheaper AI wins.
  • Nvidia stopped shipping its H200 AI chips to China, days after Jensen Huang's resignation.
  • OpenAI's models improved AMD's flagship GPU design by up to 14% — and some of those changes are already headed to manufacturing.

1. Anthropic's IPO Is the Biggest Ever — and It Starts Trading Today

The Anthropic IPO is no longer a rumor or a filing: it is the largest initial public offering in history. According to CNBC, the Claude maker priced its shares at $83 on Wednesday, above the $70 to $78 range it had marketed, and sold 980 million shares to raise about $81.3 billion — topping SpaceX's $75 billion debut from June. Trading begins today, Thursday, November 19, on the Nasdaq under the ticker ANT.

At $83 a share, Anthropic is valued at roughly $1.5 trillion on a fully diluted basis. Demand was so strong that the company enlarged the offering from 825 million shares, and a separate $26 billion private placement brings in Google, Nvidia, AMD, Broadcom, Microsoft, Micron, and TSMC as investors. Anthropic reported $33.24 billion in revenue for the first nine months of 2026, up from $2.55 billion a year earlier. This is also the first time a major frontier AI lab has gone public — the market finally gets a daily verdict on an AI company instead of a private funding round every few months. We covered the run-up in yesterday's roundup.

2. Google Ships Gemini 4 Neon, Its Workhorse Model

Google is not letting the IPO steal all the attention. According to Google, the company launched Gemini 4 Neon on November 18, the fourth model in its Gemini 4 family. Neon sits in the middle of the lineup: smarter than the fast, cheap Hydrogen model and more affordable than the heavyweight Carbon.

The pitch is about finishing jobs, not just starting them. Google describes Neon as built for "frontier-level intelligence with fast end-to-end task completion" across complex software engineering, knowledge work, and science. It scores 61 on the Artificial Analysis Intelligence Index — third overall — and tops the Coding Agent Index with a score of 58.6. Pricing is $3 per million input tokens and $15 per million output tokens, with a one-million-token context window. Neon is rolling out now in the Gemini API through Google AI Studio, in Gemini for business, in Google Antigravity for paid AI Pro and Ultra subscribers, and in the Gemini app for Pro and Ultra users. One quirk: Carbon is still on promotional pricing of $2 input / $10 output through November 30, so Neon is briefly the more expensive option.

3. Microsoft Bets That Cheap AI Beats Smart AI

At its Ignite conference in San Francisco on November 17, Microsoft made its clearest move yet to compete with the AI labs it has long relied on. According to Microsoft, the company launched seven in-house MAI models in one go, covering coding, cybersecurity, image generation, live speech, transcription, and model routing.

The headline act is MAI Code 2, now available in GitHub Copilot. Microsoft says it delivers 98% of OpenAI's GPT-6.1 Sol coding performance at about 40% of the cost, and 96% of Anthropic's Claude Opus 5.5 at just 5% of the cost. Those are Microsoft's own numbers, run on its own evaluation platform, so take the exact percentages with a grain of salt — but the direction is unmistakable: Microsoft wants to win on price-performance, not benchmark bragging rights. The other big Ignite announcement was Copilot Autopilot, an always-on agent with its own identity that watches Teams channels and mailboxes and handles recurring work without being asked. Microsoft is turning Copilot from a chatbot you prompt into a colleague that works while you are away.

4. Nvidia Stops Shipping H200 Chips to China

Nvidia has stopped shipping its H200 AI chips to China, closing its only US-approved sales channel there. According to Reuters, interim CEO Colette Kress confirmed the halt on the company's earnings call this week, saying "we are no longer shipping licensed Hopper data center products to China."

The move comes just days after longtime CEO Jensen Huang resigned on November 12, and it follows new Commerce Department rules targeting the "remote access loophole" — Chinese firms renting compute from data centers in other countries. The halted sales were worth $2.4 billion in the quarter ended October 25, about 2% of Nvidia's $115.8 billion in quarterly revenue, and the company's current guidance assumes none of it. Kress gave no reason for the halt and did not say whether it is permanent, but Congress has scheduled interviews with Huang for December 1, and scrutiny of how restricted chips reached China is only intensifying. For the full backstory on Huang's exit, see our November 13 roundup.

5. OpenAI's Models Are Now Designing the Chips They Run On

In the most quietly remarkable story of the week, OpenAI says its models are now doing chip design work good enough to make it into real silicon. According to OpenAI, its models rewrote parts of the SystemVerilog RTL — the code that describes a chip's circuits — for AMD's flagship GPU IP, delivering up to 14% improvement in power, performance, and area against baselines that were already heavily optimized. OpenAI calls that roughly a year of design progress in under five months.

AMD has already included some of the model-proposed changes in a design sent for manufacturing and added others to production design files for future products. The same approach cut standard-cell area by about 13% in the vector unit and 10% in the matrix unit of OpenAI's own Jalapeño inference chip — proven in actual silicon, not just simulation. There are caveats: OpenAI names no AMD product, gives no breakdown of the 14%, and AMD has published no statement of its own. But OpenAI is also partnering with Synopsys to train GPT-Synopsys, a model that uses professional chip-design tools the way an expert engineer would. The feedback loop is now literal: AI designs better chips, which run better AI, which designs better chips.


The Takeaway

Five stories, one thread: AI is becoming infrastructure, and the fight is moving from who has the smartest model to who controls the economics around it. Anthropic is now a public company with a $1.5 trillion price tag, Google and Microsoft are competing on cost per finished task, Nvidia is rethinking where its chips can go, and OpenAI's models are quietly redesigning the hardware underneath it all. The model wars are not over — but the money wars have officially begun.

Check back tomorrow for the next roundup.


FAQ

When does Anthropic stock start trading?

Anthropic shares begin trading on the Nasdaq under the ticker ANT on Thursday, November 19, 2026, after pricing the IPO at $83 a share.

How big was Anthropic's IPO?

At $81.3 billion raised across 980 million shares, it is the largest IPO on record, topping SpaceX's $75 billion debut in June 2026. Anthropic's fully diluted valuation at the IPO price is about $1.5 trillion.

What is Google Gemini 4 Neon?

Gemini 4 Neon is Google's mid-tier Gemini 4 model, launched November 18, 2026, built for complex software engineering, knowledge work, and science. It costs $3 per million input tokens and $15 per million output tokens, with a one-million-token context window.

Why did Nvidia stop shipping H200 chips to China?

Interim CEO Colette Kress confirmed the halt on Nvidia's earnings call, days after Jensen Huang's resignation and alongside new US Commerce Department export rules. Kress gave no reason for the decision and did not say whether it is permanent.


Sources

Leave a Reply

Your email address will not be published. Required fields are marked *