Demo: Cloud Summit 2027 is a fictional conference made to show the legit2cloud conference platform. Its people, companies and sessions are invented.

Schedule

All times are local to the venue (America/Chicago).

Day 1 · Wednesday, March 17

  1. 9:30 AM – 10:15 AM
    Main Hall
    KeynoteAI Media Pipelines

    From Hours of Footage to Minutes of Story

    Adaeze Okafor walks through two years of running AI clipping, captioning and dubbing for news video at Northbeam Labs: what the pipeline looks like, where humans still review every output, and the three failures that taught the team the most.

  2. 10:45 AM – 11:30 AM
    Room A
    TalkGPU Economicsintermediate

    What a GPU-Hour Really Costs

    Tolu Adeyemi breaks a GPU-hour into power, depreciation, idle time and support, then compares datacenter cards with community-owned consumer cards for real media workloads such as transcription, image generation and short video renders.

  3. 10:45 AM – 11:30 AM
    Room B
    TalkCloud Infrastructureintermediate

    Object Storage for a Million Clips

    Siobhan Brennan explains how Quarrystone Data stores and serves millions of short clips: lifecycle rules, when to delete, and how retention policies interact with users who expect their files to stay forever.

  4. 11:45 AM – 12:30 PM
    Room A
    TalkTrust & Safety

    Label It Honestly: AI Disclosure That Readers Trust

    Marisol Reyes shares Bluefen Media's labelling rules for AI-assisted articles, images and audio, the reader research behind them, and what changed in complaints and corrections after the labels went live.

  5. 11:45 AM – 12:30 PM
    Room B
    TalkAI Media Pipelinesadvanced

    Dubbing Without the Uncanny Valley

    Kenji Nakamura shows how Oak & Lumen Studios aligns translated speech with the original timing, which voices are cleared for use, and how the team checks results before anything ships.

  6. 12:30 PM – 1:30 PM
    Exhibit Hall
    Break

    Lunch

  7. 1:30 PM – 3:30 PM
    Workshop Lab
    WorkshopAI Media Pipelinesintermediate

    Workshop: Build a Captioned Clip Pipeline

    In this hands-on lab, Diego Garcia and Kenji Nakamura guide attendees through building a small pipeline that transcribes a recording, picks highlights, and renders captioned vertical clips. Bring a laptop; GPU time is provided.

  8. 1:30 PM – 2:15 PM
    Room A
    TalkCloud Infrastructureadvanced

    Moving Model Weights Across Regions

    Kwame Mensah covers how Copperline Networks moves tens of gigabytes of model weights between regions without saturating shared links, including caching, scheduling transfers off-peak, and what happens when a pull goes wrong.

  9. 3:45 PM – 4:45 PM
    Main Hall
    PanelTrust & Safety

    Panel: AI in the Small Newsroom

    Adaeze Okafor (moderator), Lena Fischer (panelist), Ewa Kowalski (panelist), Marisol Reyes (panelist)

    Lena Fischer, Ewa Kowalski and Marisol Reyes discuss AI-written briefs, podcasts and illustrations in small newsrooms: what they publish, what they refuse to automate, and how they keep a human editor accountable. Moderated by Adaeze Okafor.

  10. 5:00 PM – 7:00 PM
    Exhibit Hall
    Social

    Evening Reception

Day 2 · Thursday, March 18

  1. 9:30 AM – 10:15 AM
    Main Hall
    KeynoteGPU Economics

    Ten Thousand Cards, One Queue

    Henrik Lindqvist explains how Tessellate Cloud schedules work across thousands of mixed consumer and datacenter GPUs: matching jobs to memory, handling machines that disappear mid-task, and keeping paying work first.

  2. 10:45 AM – 11:30 AM
    Room A
    TalkTrust & Safetyadvanced

    Provenance Signals That Survive Re-uploads

    Omar Haddad presents Ardentide research on which provenance signals survive cropping, compression and re-uploads to social platforms, and which ones quietly disappear.

  3. 10:45 AM – 11:30 AM
    Room B
    TalkAI Media Pipelinesbeginner

    Image Generation for Marketing Teams

    Ritu Patel shows how Ferrowind AI customers generate campaign imagery: style guides as prompts, review queues, and the guardrails that keep real people and trademarks out of the output.

  4. 11:45 AM – 12:30 PM
    Room A
    TalkCloud Infrastructureintermediate

    When Half Your Fleet Is Someone's Gaming PC

    Diego Garcia shares Gridhollow's playbook for running inference on donated and rented consumer machines: health checks, safe retries, and why a node that looks online may not be doing any work.

  5. 12:30 PM – 1:30 PM
    Exhibit Hall
    Break

    Lunch

  6. 1:30 PM – 2:30 PM
    Main Hall
    Lightning talksAI Media Pipelines

    Lightning Talks: Shipped It

    Short talks from teams that shipped AI media features in the past year, including an AI-assisted weekly podcast from Ewa Kowalski and a business-briefing experiment from Lena Fischer.

  7. 2:45 PM – 3:30 PM
    Room A
    TalkGPU Economicsintermediate

    Budgeting AI Media Without Surprises

    Tolu Adeyemi and Henrik Lindqvist walk through forecasting compute spend for products that generate audio, images and video on demand, including caps, credits and what to do when usage spikes.