Oracle to Offer Cerebras AI Inference on Cloud in 2027

Cerebras stock jumped 10% on October 27 after Oracle announced it will offer Cerebras AI inference on its cloud starting in 2027. The deal gives Cerebras a second sales channel beyond OpenAI. Oracle also revealed partnerships with Google TPUs, SambaNova, and NVIDIA, though neither company disclosed dollar figures for the agreement.

Meta's announcement that it plans to raise up to $95 billion in stock to fund its AI infrastructure build-out sent Vicor shares up 11%. Vicor reported revenue of $197.1 million, up 78.5% year over year, and analysts expect its revenue to more than double from 2025 to 2027. Despite the strong growth, the stock still trades below its five-year average forward P/E ratio.

Alphabet shares fell 8% in after-hours trading even as Google Cloud sales doubled to $30.7 billion. The company spent $51.4 billion on data centers and capital projects in the quarter, pushing free cash flow to minus $9.1 billion. Total revenue rose 25% to $127.6 billion but missed analyst expectations, and capital spending jumped 114% from a year earlier.

AI security is struggling to keep pace with fast technology adoption. A report found that 67% of decision-makers in Singapore felt pressured to approve AI deployments despite known security concerns, while only 29% felt prepared for the speed of adoption. The report calls for security to be embedded from day one rather than added after deployment.

Jedify launched Data Agents, a tool that lets users build AI analysts by describing questions in plain language. The system uses a context graph mapping a company's data, metrics, and permissions so agents can answer questions without raw schema exploration. Results respect individual user permissions and can be delivered via email or Slack. Jedify has raised over $33 million in venture capital funding.

The fully AI-generated feature film Sanxingdui: Future Past earned only about $181,000 in its opening weekend in Chinese theaters. The film had a budget of $1.5 million and was produced over two years by a team of 70 to 80 people. It is the first AI-generated feature to receive permission for screening in Chinese theaters, but its poor performance raises questions about audience expectations for cinematic quality.

Haivision's 2026 Public Safety Technology Report found that 77% of public safety professionals use fixed surveillance cameras and 57% use drones. Seventy-six percent believe AI-powered video analytics will have the greatest impact on public safety over the next five years, though challenges like budget constraints and integration issues remain.

A McKinsey Global Institute study found AI had limited impact on U.S. productivity growth over the past decade. AI boosted productivity in manufacturing and logistics but had less effect in services and finance, and the benefits have been concentrated among a small group of workers rather than spread across the workforce.

NYU researchers developed a physics-informed AI system that forecasts wildfire spread in near real time by combining a fire-spread model with an AI layer. The tool was tested on three wildfires including Oregon's Grasshopper Fire, achieving a 94.6% match between predicted and observed fire areas. The team has filed a provisional U.S. patent.

CASEL, a new runtime system, decides how carefully an AI agent should check each tool call before acting. In tests with 120,000 actions, CASEL cut harmful side effects from 14.75 percent to 7.02 percent while using the same compute cost as other methods. The approach is model-agnostic and works with any tool-using language agent.

Liquid AI released the LFM3 family on October 27, offering six open-weight models built on one shared backbone. The lineup includes models for text, vision, audio, and all modalities. LFM3-VL-2B raised vision scores from 57.6 to 71.4, and all models can be fine-tuned on a single GPU. Liquid AI also adapted LFM3-VL-2B into three robotics models.

Key Takeaways

  • Cerebras stock rose 10% after Oracle announced it will offer Cerebras AI inference on its cloud starting in 2027, giving Cerebras a second sales channel beyond OpenAI.
  • Meta plans to raise up to $95 billion in stock to fund its AI infrastructure build-out, while Vicor reported revenue of $197.1 million, up 78.5% year over year.
  • Alphabet shares fell 8% despite Google Cloud sales doubling to $30.7 billion, as $51.4 billion in capital spending pushed free cash flow to minus $9.1 billion.
  • A Singapore report found 67% of decision-makers felt pressured to approve AI deployments despite security concerns, and only 29% felt prepared for the pace of adoption.
  • Jedify launched Data Agents, a plain-language AI analyst tool that maps company data and permissions, and has raised over $33 million in venture funding.
  • The first AI-generated feature film to screen in Chinese theaters, Sanxingdui: Future Past, earned only about $181,000 in its opening weekend on a $1.5 million budget.
  • Haivision's report found 76% of public safety professionals believe AI-powered video analytics will have the greatest impact on public safety over the next five years.
  • A McKinsey Global Institute study found AI had limited impact on U.S. productivity growth over the past decade, with benefits concentrated among a small group of workers.
  • NYU researchers developed a physics-informed AI system that predicts wildfire spread in near real time, achieving 94.6% accuracy on Oregon's Grasshopper Fire.
  • CASEL reduced harmful AI agent side effects from 14.75% to 7.02% in testing while using the same compute cost, and Liquid AI launched six LFM3 models covering text, vision, audio, and omni modalities.

Cerebras stock jumps 10% on Oracle cloud AI deal

Cerebras shares rose 10% on October 27 after Oracle announced it will offer Cerebras AI inference on its cloud starting in 2027. Oracle also revealed deals with Google TPUs, SambaNova, and NVIDIA. The deal gives Cerebras a second sales channel beyond OpenAI. Neither company disclosed dollar figures for the deal.

Vicor stock rises on Meta's $95 billion AI funding plan

Vicor shares jumped 11% after Meta announced plans to raise up to $95 billion in stock to fund its AI infrastructure build-out. Vicor reported revenue of $197.1 million, up 78.5% from a year earlier. Analysts expect Vicor revenue to more than double from 2025 to 2027. The stock still trades below its five-year average forward P/E ratio.

Alphabet stock drops 8% despite doubling cloud revenue

Alphabet shares fell 8% in after-hours trading on Wednesday even as Google Cloud sales doubled to $30.7 billion. The company spent $51.4 billion on data centers and capital projects in the quarter, pushing free cash flow to minus $9.1 billion. Total revenue rose 25% to $127.6 billion but missed analyst expectations. Capital spending jumped 114% from a year earlier.

Alphabet falls 8% as AI spending outpaces cash flow

Alphabet shares dropped 8% after hours as spending of $51.4 billion pushed quarterly free cash flow to minus $9.1 billion. Google Cloud sales doubled to $30.7 billion and profit margins widened to 36.2%. Fed Chair Kevin Warsh said he doubts AI is safe from rising interest rates. Alphabet and four peers now make up 4.8% of the US investment-grade corporate bond market.

AI security must keep pace with fast technology adoption

A report warns that AI is moving faster than the security built to protect it. In Singapore, 67% of decision-makers said they felt pressured to approve AI deployments despite known security concerns. Only 29% felt prepared for the pace of AI adoption. The report calls for security to be embedded from day one rather than added after deployment.

Jedify launches Data Agents for AI-powered business analysis

Jedify launched Data Agents, a tool that lets users build AI analysts by describing questions in plain language. The system uses a context graph that maps a company's data, metrics, and permissions so agents can answer questions without raw schema exploration. Agents run on request, on a schedule, or when a user-defined condition is met. Results respect individual user permissions and can be delivered via email or Slack. RubyPlay is already using the product for daily automated reports. Jedify has raised over $33 million in venture capital funding.

AI-generated film Sanxingdui flops at Chinese box office

The fully AI-generated feature film Sanxingdui: Future Past earned only about $181,000 in its opening weekend in Chinese theaters. The film had a budget of $1.5 million and was produced over two years by a team of 70 to 80 people. It is the first AI-generated feature to receive permission for screening in Chinese theaters. The film originated as an AI microdrama series that reached nearly 200 million views. Bona Film Group has already begun production on a second AI-generated feature. The film's poor theatrical performance raises questions about audience expectations for cinematic quality.

Haivision report shows rising AI investment in public safety

Haivision's 2026 Public Safety Technology Report surveyed over 100 public safety professionals. It found that 77% use fixed surveillance cameras, 57% use drones, and 54% use body-worn cameras. Forty-five percent plan to invest in Drone as First Responder programs over the next 12 months. Seventy-six percent believe AI-powered video analytics will have the greatest impact on public safety over the next five years. The report highlights challenges including budget constraints, integration issues, and information-sharing across agencies.

AI has limited impact on U.S. productivity growth so far

The U.S. saw its strongest productivity surge in decades with a 2.5% rise in the third quarter, but the increase was largely due to fewer hours worked rather than higher output per hour. A McKinsey Global Institute study found AI had limited impact on productivity growth over the past decade. AI boosted productivity in manufacturing and logistics but had less effect in services and finance. The benefits have been concentrated among a small group of workers rather than spread across the workforce. A 2020 Federal Reserve study reached similar conclusions about AI's limited productivity gains.

Physics-informed AI tool predicts wildfire spread in real time

NYU researchers developed a physics-informed AI system that forecasts wildfire spread in near real time by combining a fire-spread model with an AI layer. The tool continuously updates using live data from satellites, weather feeds, drones, ground sensors, and field crews. Users can select a location, area size, and forecast horizon up to 12 hours on an interactive map. An AI chatbot interface lets users ask questions about containment strategies and evacuation plans. The system was tested on three wildfires including Oregon's Grasshopper Fire, achieving a 94.6% match between predicted and observed fire areas. The team has filed a provisional U.S. patent.

CASEL improves safety for AI tool use with selective execution

CASEL is a new runtime system that decides how carefully an AI agent should check each tool call before acting. It uses factors like failure probability, consequence severity, reversibility, and evidence freshness to choose between direct execution, verification, sandboxing, or human approval. In tests with 120,000 actions per seed across 20 seeds, CASEL cut harmful side effects from 14.75 percent to 7.02 percent. It also lowered mean normalized harm from 0.184 to 0.123 while using the same compute cost as other methods. The approach is model-agnostic and works with any tool-using language agent.

Liquid AI launches LFM3 models with text, vision, audio and omni support

Liquid AI released the LFM3 family on October 27, 2026, offering six open-weight models built on one shared backbone. The lineup includes LFM3-1.7B for text, LFM3-VL-2B for text and images, LFM3-Audio-2B for speech, LFM3-Omni-2B for all modalities, and two decision models called d1-2B and d1-omni-2B. The backbone uses 24 layers combining gated convolutions, sliding-window attention, and global attention with innovations like head-wise gated attention and partial RoPE. Liquid AI reports strong benchmark gains, such as LFM3-VL-2B raising vision scores from 57.6 to 71.4 and LFM3-Audio-2B supporting 18 languages with speech quality of 3.83 on UTMOSv2. All models can be fine-tuned on a single GPU, and Liquid AI also adapted LFM3-VL-2B into three robotics models.

Sources

NOTE:

This news brief was generated using AI technology (including, but not limited to, Google Gemini API, Llama, Grok, and Mistral) from aggregated news articles, with minimal to no human editing/review. It is provided for informational purposes only and may contain inaccuracies or biases. This is not financial, investment, or professional advice. If you have any questions or concerns, please verify all information with the linked original articles in the Sources section below.

AI Stocks Cloud Computing AI Partnerships AI Security AI Adoption AI Film AI in Public Safety AI Productivity AI Wildfire Prediction AI Agent Safety AI Multimodal Models

Comments

Loading...