AI news explained in motion graphics
How to start a faceless AI news YouTube channel
The week's biggest AI story (a new model, chip or deal) explained simply: what happened, how it works, why it matters.
How Cliptude makes it: Faceless narration + sourced slides on a dark slate, trend charts, a globe, a status on every claim
The reference channel: WhatsWithAI
@WhatsWithAInow · joined 2026-02-25 · checked 14 November 2026
Why this niche works
- WhatsWithAI went from almost nothing to 992,344 views and 5,040 subscribers in about a month, after "DeepSeek Just Shook the Entire AI Industry Again" reached 816K views.
- estimates put it at 1,265,486 monthly views and a $3.08 RPM, about $3,898 a month of potential revenue; it is not monetized yet but is past 1,000 subscribers.
- Its next two videos (open models on PS5, 83K; NVIDIA chip gains, 69K) show the audience stayed for news explainers, not just one viral hit.
- Monetized peers prove the money: KernelBrief (31 days) earns about $1,419 a month on local-AI explainers, and the older TheAIGRID makes about $15,488 a month on daily AI news.
Top videos on WhatsWithAI




What a copy could earn
Videos a month × views per video × $3.08 RPM. Illustrations, not promises.
| Slow start | 8 × 4,000 | $99/mo |
| On pace | 12 × 30,000 | $1,109/mo |
| Breakout | 12 × 150,000 | $5,544/mo |
Competition
HighAI news is crowded with big daily channels (TheAIGRID 407K subs, AI Revolution 577K) and RPMs are modest ($2.5-$6). New channels break in by explaining one big story clearly and visually within days of it happening, rather than posting a daily roundup.
| Channel | Subs | Avg views |
|---|---|---|
| TheAIGRID 1,395 days, daily AI news, RPM $6.20, ~$15,488/mo |
407,000 | 51,433 |
| KernelBrief 31 days, local-AI hardware explainers, monetized, ~$1,419/mo |
2,450 | 10,482 |
| Understanding AI 61 days, animated concept series, not monetized yet |
44,300 | 6,607 |
How the Claim Check Look matches
WhatsWithAI uses clean dark motion graphics: a near-black grid background, rounded cards with teal and white outlines, short labels like READ vs WRITE and "PURE MATH POWER", and simple comparisons instead of footage. Claim Check tells a news story in a similar dark, card-led language: one claim per Briefing slide on a dark slate, tagged established, disputed or unknown, with a big number and its source line, two to four values as bars, trends on Line chart callout and places on Night globe focus. It does not draw glowing pipelines or icons, so describe each process as plain steps across slides, and it will not show live product demos or screen recordings of a new model.
Make it in Cliptude
- 1Pick the Claim Check Look, format Faceless documentary, 8-12 minutes.
- 2Script shape: the news in one line, what actually happened, how it works (one process, step by step), the numbers, who wins and loses, and what to watch next.
- 3In the prompt give each figure with its source, say which claims are confirmed and which are forecasts, and list the trends you want on Line chart callout.
- 4Make it within 48 hours of the news; date-check every number and name the source in the script.
- 5Thumbnail reference: WhatsWithAI's DeepSeek video, one company logo and a short shock line on a dark background.
- 6Post 2-3 a week around real news; avoid filler weeks, the reference's views dropped to 116,158 in 14 days when it slowed down.
AI news video title ideas
The formula WhatsWithAI wins with: [Company] Just [did something] (fill in with this week's story), or Why [trend] Is Happening
- [Company] Just Released [model]: What Actually Changed
- Why AI Data Centers Need Their Own Power Plants
- What an AI Chip Actually Does, Explained in 8 Minutes
- Why Every Phone Maker Is Suddenly Adding AI Chips
- [Company] vs [Company]: Who Is Really Winning the AI Race This Month
- The Real Cost of Training an AI Model
Fresh titles in the same pattern. Check each story's facts before you script it.
Why AI Companies Are Suddenly Buying Nuclear Power
It is one of the clearest 'AI meets the real world' stories, it keeps producing new deals, and every step fits a simple energy pipeline.
The strange headline
In September 2024, Microsoft signed a deal to help bring back a nuclear reactor at Three Mile Island, the most famous name in American nuclear history. Not to sell electricity. To buy it, every hour, for twenty years. Since then Meta, Google and Amazon have all signed nuclear deals of their own. Why are software companies, businesses built on code, suddenly shopping for reactors? The answer is the thing every AI model quietly runs on. Not data. Not chips. Electricity. Lots of it, all day and all night, without a single break. Let's follow the power from the plant to your chatbot.
On screen: Briefing slide 'Microsoft + Three Mile Island' with the kicker THE DEAL, status established and its source line; Night globe focus on Pennsylvania.
How much power AI needs
Every question you ask an AI model runs on thousands of chips in a data centre, and those chips turn electricity into heat while they work. Then more electricity is spent cooling them. According to the International Energy Agency, data centres worldwide used about 415 terawatt hours of electricity in 2024. That is roughly one and a half percent of all the electricity on Earth. By 2030, the agency expects that to more than double, to around 945 terawatt hours, slightly more than all of Japan uses today. And in the United States, data centres are expected to make up nearly half of all growth in electricity demand this decade.
On screen: Briefing slide bars '415 TWh (2024)' and '945 TWh (2030)', the 2030 forecast tagged unknown; Briefing slide '~1.5% of world electricity' with its source; Line chart callout of data-centre demand to 2030.
Why nuclear
So why nuclear, and not just more solar panels and wind farms? Because a data centre never sleeps. Its chips need the same steady power at three in the morning as at noon, on calm days and cloudy days. Solar and wind are cheap, but they rise and fall with the weather, so they need large batteries or backup to run around the clock. A nuclear reactor, once running, produces a steady flow of electricity day and night for decades, and it does not burn fossil fuel to do it. For tech companies that have promised to cut their carbon emissions, that combination is almost exactly what they want.
On screen: Two slides: 'Solar and wind rise and fall with the weather' and 'Nuclear runs day and night', status established.
The Three Mile Island deal
Here is how the Microsoft deal works. Three Mile Island had two reactors. Unit 2 was damaged in the famous accident of 1979. Unit 1 kept running safely for decades until it closed in 2019 because it was losing money. Constellation, the energy company that owns it, agreed to restart Unit 1 under a new name, the Crane Clean Energy Center. It plans to spend about 1.6 billion dollars to bring roughly 835 megawatts back onto the grid, with a target of 2028. Microsoft signed a 20-year agreement to buy power matching what its data centres use. An old reactor gets a second life because of AI.
On screen: Briefing slide 'Unit 1: closed 2019', then '$1.6B restart', '835 MW', '20-year deal' and 'Target: 2028' tagged unknown; a separate slide 'Unit 2: damaged in the 1979 accident'.
Everyone joins in
Microsoft was not alone for long. In June 2025 Meta signed its own 20-year deal with Constellation to buy power from a nuclear plant in Illinois. Google made the first corporate agreement to buy power from a fleet of small modular reactors being developed by a company called Kairos Power, and in October 2026 it agreed to buy about 890 megawatts from upgrades at existing Constellation plants. Each deal works the same way. The tech company promises to buy power for many years, and that guaranteed buyer helps the energy company justify spending billions to keep, upgrade or build reactors.
On screen: Briefing slide bars of the deals: Microsoft 835 MW, Google 890 MW of uprates; slides for Meta's 20-year Illinois deal and Google's Kairos small reactors; Night globe focus on Illinois.
The catch
There are real limits. New reactors take many years to license and build, and small modular reactors are still mostly plans, not running plants. Restarting or upgrading old reactors is faster, but there are only so many of them. The International Energy Agency itself warns that its forecasts for AI electricity demand are highly uncertain, because chips keep getting more efficient and nobody knows exactly how fast AI use will grow. So these deals are partly a bet. Tech companies are locking in steady power now, before everyone else needs it, in case demand grows as fast as they expect.
On screen: Briefing slide 'New reactors take years to build', status established; 'Small modular reactors: mostly planned', status unknown; Line chart callout of two possible demand curves, labelled as forecasts.
What it means
So the next time you type a question into a chatbot, picture the chain behind the answer. A server rack, a cooling system, a power line, and somewhere at the end of it, maybe, a reactor that was closed a few years ago and is coming back because of you. AI is not just a software story anymore. It is an energy story. If you want the next big AI story explained this simply, subscribe, and tell us what you want broken down next.
On screen: Final Briefing slide 'Reactor, grid, data centre, your answer'; end card on the dark slate.
AI news channel questions
How many views does a AI news channel need to make $2,000 a month?
At the $3.08 RPM estimated for WhatsWithAI, about 649,351 long-form views a month make $2,000 from AdSense. That is roughly 14 videos at this channel's average of 45,034 views each.
How much does a AI news YouTube channel make?
WhatsWithAI is 262 days old with 5,040 subscribers and earns an estimated $3,898 a month once monetized from 1,265,486 monthly views. Sponsorships and affiliate income come on top of that.
Is AI news a good YouTube niche in 2026?
Competition is high. AI news is crowded with big daily channels (TheAIGRID 407K subs, AI Revolution 577K) and RPMs are modest ($2.5-$6). New channels break in by explaining one big story clearly and visually within days of it happening, rather than posting a daily roundup. Channels to study: TheAIGRID, KernelBrief, Understanding AI.
Can you make AI news videos without showing your face?
Yes. In Cliptude, the Claim Check Look makes these videos from a script, with no presenter on camera: faceless narration, sourced slides on a dark slate, trend charts, a globe, a status on every claim.
Make your first video in this niche
Open the Claim Check Look, paste the script above or your own topic, and Cliptude builds the narration, footage and on-screen boards.
Sources
- https://www.youtube.com/channel/UCxg7h82QQ46A4jo0ejg8Vgg/videos
- https://socialblade.com/youtube/channel/UCxg7h82QQ46A4jo0ejg8Vgg
- https://www.iea.org/reports/energy-and-ai/executive-summary
- https://www.constellationenergy.com/newsroom/2024/Constellation-to-Launch-Crane-Clean-Energy-Center-Restoring-Jobs-and-Carbon-Free-Power-to-The-Grid
- https://www.utilitydive.com/news/constellation-three-mile-island-nuclear-power-plant-microsoft-data-center-ppa/727652/
- https://insideclimatenews.org/news/03062025/meta-constellation-nuclear-power-deal/
- https://thedailyrecord.com/2026/10/06/md-based-constellation-power-deal-google/
Figures are estimates checked 14 November 2026, cross-checked on Social Blade and YouTube. Revenue is AdSense only and real RPM varies. Channel names and video titles belong to their owners; Cliptude is not connected to any channel listed. All 26 niches