Home › News › AI Daily October 28 Noon Edition

AI Daily | October 28, 2026 Noon Edition: Anthropic Files for the Largest IPO in History, OpenAI's GPT-6.1 Terra Price Halved

Published: 2026-10-28 12:00  |  Category: AI Daily  |  Read: about 7 min

Top Stories This Noon

  • Anthropic formally files its IPO prospectus: planning a Nasdaq listing under the ticker ANT, at a valuation of up to $2 trillion and raising about $100 billion
  • OpenAI releases GPT-6.1 Terra: API prices 50% below the previous generation, at $1 per million tokens for input and $5 for output
  • SpaceX and Apollo reach a $40 billion GPU financing deal: funding support for AI compute expansion
  • iFlytek launches SparkAudio 1.0: a speech foundation model trained on fully domestic compute
  • Hinton warns about AI risk: puts the probability of superintelligence wiping out humanity at about 11.73%

1. Anthropic Formally Files Its IPO Prospectus: the Largest AI Company IPO Ever Is Here

AnthropicIPONasdaq

On October 26 local time, Anthropic publicly submitted its S-1 prospectus to the US Securities and Exchange Commission (SEC), planning to list on the Nasdaq Global Select Market under the ticker "ANT", with Goldman Sachs leading the underwriting.

  • The prospectus shows first-half 2026 revenue of $16.3 billion, with Q2 already producing $974 million in operating profit
  • Signed compute contracts total about $657 billion, covering Google, Amazon, Microsoft, SpaceX and others
  • Plans to raise about $100 billion, at a valuation of up to $2 trillion
  • The filing devotes 93 pages to AI risk factors, warning that models could extort users or even threaten human survival
Editor's note: This is the first time Wall Street has publicly priced a frontier AI model company. For years everyone has asked "how much is a large model actually worth"; now there is finally an answer that carries legal liability. Anthropic put "safety" in its brand name; whether its safety narrative can withstand the test of quarterly earnings in the capital markets is worth watching closely.

2. OpenAI Releases GPT-6.1 Terra, Halving API Prices

OpenAIGPT-6.1 TerraPrice Cut

On the same day Anthropic filed, OpenAI announced GPT-6.1 Terra, cutting standard API prices by 50% versus the previous generation, GPT-5.6 Terra.

  • Standard API pricing: $1 per million input tokens and $5 per million output tokens, with cached input as low as $0.05
  • Context window reaches 1.1M tokens
  • Focuses on visual understanding and computer operation, handling tasks such as filling in forms, editing video and reading technical drawings
  • Rolled out to Plus, Pro, Enterprise and Edu users in ChatGPT Work and Codex
Industry impact: This is not a simple discount promotion. When a model becomes an order of magnitude cheaper, the competitive focus shifts from the model layer to the workflow layer. Developer tools such as Greptile and CodeRabbit have already integrated it, intensifying competition in AI coding.

3. SpaceX and Apollo Strike a $40 Billion GPU Financing Deal as the Compute Arms Race Escalates

SpaceXApolloGPU Financing

SpaceX and Apollo have reached a $40 billion GPU financing arrangement to fund its rapidly expanding AI compute infrastructure.

  • The financing structure lets SpaceX avoid relying entirely on traditional debt markets
  • Reflects that AI compute demand is still growing exponentially
  • Global AI investment is expected to reach $1 trillion by year-end
  • The EU's "InvestAI" plan aims to mobilize 200 billion euros for compute infrastructure
Risk note: In its latest report the IMF lists an "AI-driven market correction" as one of three near-term risks, and the payback period for trillion-dollar compute investments remains highly uncertain.

4. iFlytek Releases SparkAudio 1.0, a Speech Model Trained on Fully Domestic Compute

iFlytekSparkAudioDomestic Compute

iFlytek has officially released the speech foundation model SparkAudio 1.0, trained on fully domestic compute, marking a new stage for Chinese speech AI.

  • Uses three core technologies—SemiEnc, S-S2S and CoDec—to strengthen representation, cut latency and reduce compute overhead
  • Spins off three dedicated models including Spark-ASR-2.0, raising simultaneous interpretation MOS from 4.4 to 4.6
  • Already deployed in iFlytek's input method, AI recorder ring and other scenarios
  • A 35-gram recorder ring enables all-day AI voice interaction

5. Hinton Warns: About an 11.73% Chance Superintelligence Wipes Out Humanity

HintonAI SafetySuperintelligence

The "godfather of AI", Geoffrey Hinton, delivered a major speech at the 2026 Vienna Gödel Lecture, issuing a stern warning about the risks of superintelligence.

  • He estimates the probability of superintelligence destroying humanity at 10%-20%, with a precise figure of about 11.73%
  • Notes that the human brain runs on just 20 watts but communicates at about 10 bits per second, while digital models can synchronize knowledge instantly through gradient averaging
  • Argues for a "maternal instinct" that makes AI place human wellbeing above its own
  • Reveals that he left Google mainly because he wanted the freedom to speak out about AI risk
Safety warning: With Anthropic listing AI risks across 93 pages of its IPO prospectus and Hinton offering a precise extinction probability, the whole industry is going through a cognitive shift from "technological optimism" to "responsible development."

6. Berkeley Professor Drops Graded Homework Over AI Cheating

EducationBerkeleyAI Cheating

A professor at the University of California, Berkeley has announced that he will no longer assign graded homework because of worsening AI cheating.

  • The professor instead favors traditional assessments such as in-person exams
  • AI is reshaping educational assessment at unprecedented speed
  • Many universities are exploring new approaches to academic integrity in the AI era
  • It has sparked broad debate about "the nature of education in the AI era"

7. Broadcom Secures $63 Billion in Financing to Supply AI Chips to Anthropic

BroadcomChipsFinancing

Chip company Broadcom has reached a $63 billion financing deal with Bank of America, Blackstone, Citi and Morgan Stanley, mainly to supply custom AI chips to Anthropic.

  • One of the largest financings in the AI chip sector to date
  • Reflects deep restructuring in the AI compute supply chain
  • Anthropic's prospectus lists a $161.2 billion contract to lease equipment from Broadcom
  • Advanced packaging prices are rising, keeping the chip supply chain tight

Today's Key Figures

MetricFigure
Anthropic IPO planned raiseAbout $100 billion
Anthropic first-half revenue$16.3 billion
Anthropic signed compute contracts$657 billion
GPT-6.1 Terra API price cut50%
GPT-6.1 Terra context window1.1M tokens
SpaceX GPU financing$40 billion
Broadcom AI chip financing$63 billion
Hinton's estimated superintelligence extinction probabilityAbout 11.73%
Global AI investment forecast by year-end$1 trillion

Sources

  • Yicai: On the same day Anthropic filed, OpenAI halved the price of its new model
  • Cailian Press: Google's Genie 4 debuts as Anthropic publicly files its Nasdaq IPO application
  • Sina AI Hotspot Hourly: AI hotspot digest, October 28, 2026
  • Tencent Research Institute: AI Express 20261028
  • PANews: Key news from last night and this morning (October 26-27)
  • AIHOT: Breaking down Anthropic's prospectus—11 details behind a planned $100 billion IPO