Everything About AI
NextFin News - Google DeepMind cut the price of its AI video-generation model by 30% on Thursday and pushed the maximum native clip length to 15 seconds, putting a 720p second of generated video at $0.07 - below OpenAI's Sora 2 Pro tier and about one-tenth of the $0.75 per second charged for top-tier 4K output. The move, announced alongside the general release of Gemini Omni 1.2 Flash, signals that the AI video market has shifted from a scarcity race for clip length into a price-and-throughput war where the cheapest usable second wins.
On the day of the announcement, Alphabet shares closed at $284.485 on Nov. 5, down $9.49, or 3.23%, in a session that ran from an open of $293.97 to an intraday low of $277.07. The decline extended a broader pullback: the shares are trading below their 50-day average of $335.23 and roughly 30% below their year-to-date high of $408.61, with a market capitalization of about $3.48 trillion. The tape delivered a cooler read than the product news, and the gap between an aggressive product move and a negative session is the question this piece pursues. Is the price cut a sign of strength, a margin giveaway, or evidence that video AI is becoming a commodity faster than investors want to admit?
What Google Actually Shipped: The Numbers Behind Omni 1.2 Flash
Gemini Omni 1.2 Flash is generally available now across the Gemini API, Google Vids, and Gemini for business, according to Anish Nangia, a product manager at Google DeepMind. Three changes define the release. First, the model can generate and edit videos natively up to 15 seconds, up from 10 seconds with Omni 1.1, with duration specifiable in one-second increments from 3 to 15 seconds. Second, Google cut the price 30% from Omni 1.1, bringing 720p output to $0.07 per second. Third, the workflow adds multi-turn conversational editing, so a creator can fix a detail with a follow-up prompt instead of regenerating the entire clip.
The pricing math matters more than the headline percentage. A 15-second 720p clip costs $1.05 at the new rate. Google's recommended workflow - draft at 360p or 720p, fix a detail with a follow-up edit, then upscale the best takes to 1080p or 4K - is designed to keep the cost of a finished spot anchored to a handful of generated seconds rather than dozens of failed attempts. That is a direct attack on the economics that have made AI video expensive to run at scale.
Reference adherence is the other substantive upgrade. Users can guide generation with character, product, or style images, short motion clips of up to three seconds, and audio cues. Google's claim is that faces, garments, packaging typography, and brand colors stay consistent across camera angles - the specific failure mode that has kept generated video out of branded advertising. The company illustrated the point with a virtual try-on sequence: one model photo plus front-and-back product shots of three outfits, rendered into a 15-second sequence across 360-degree turns and match cuts while keeping the model's identity and each garment's cut, drape, and back details consistent.
"We reduced the price by 30% to help you produce high-quality video content at a lower cost," Nangia said in the announcement.
The Price War Nobody Asked For: Why $0.07 Changes the Competitive Map
At $0.07 per second for 720p, Google is pricing below the middle of the market and within striking distance of the floor. A July 2026 survey of AI video API pricing put the market range at $0.05 to $0.75 per second of output. OpenAI's Sora 2 base tier sat at $0.10 per second, with Sora 2 Pro at $0.30 to $0.50 per second. Luma's Ray 2 ran at about $0.04 per second and Ray 3 at roughly $0.21 per second. Only xAI's Grok Imagine Video, at $0.05 per second, undercut Google's new rate - and it carries no free tier and is tied to the X platform ecosystem.
The context is that OpenAI discontinued Sora 2 in September 2026, and an estimate published at the time put the company's subsidy burn on the product at about $15 million a day for a period. That is the backdrop against which Google's cut lands: the market leader in generative AI has already retreated from subsidizing video at scale, and Google is walking into the space with a lower list price and the infrastructure to defend it.
The cost structure explains why this is a structural move, not a promotion. Eylul Kayin, a partner at Gradient Ventures, put the unit economics plainly: "At minimum, every second is 60 times more expensive than running a single image model, which is itself more expensive than voice." Video is not just text with motion; every frame in a sequence is processed, and consistency across frames multiplies the compute burden. A price cut of 30% in that environment is only possible if the underlying cost per inference has fallen - through model efficiency, better scheduling, or proprietary silicon - or if the company is willing to buy share.
Google can plausibly do both. Alphabet's most recent quarterly release, filed with the SEC, showed Google Cloud revenue of $24.8 billion in the second quarter of 2026, up 82% from a year earlier, following growth of 63% in the first quarter of 2026, 48% in the fourth quarter of 2025, 34% in the third quarter of 2025, and 32% in the second quarter of 2025. Annualized, that puts Cloud near a $99 billion run rate. Revenue from products built on Google's generative AI models grew nearly 800% year over year in the first quarter of 2026, and the Gemini app reached 950 million monthly active users by the second quarter. A business growing at that pace can absorb a price cut on one model line while making it up in volume, API lock-in, and enterprise bundling - the same playbook Google used in search and cloud infrastructure.
Why the Stock Sold Off on Good News: The Market Is Pricing a Commodity, Not a Moat
The tape told a different story from the product announcement. Alphabet shares fell 3.23% on Nov. 5, closing at $284.485, with nearly 30 million shares changing hands. The stock opened at $293.97, touched $295.60 on the high side, and spent much of the session near the $277.07 intraday low. That is a stock trading below its 50-day average of $335.23 and roughly 30% below its year-to-date high of $408.61.
The sell-off is not irrational, and reading it as such is the first mistake. Investors are not punishing Google for shipping a better model; they are repricing what AI video is worth as a business. When the defining feature of a product line becomes a lower price per unit, the market infers one of two things: either the technology is commoditizing, or the margin pool is smaller than hoped. Both conclusions compress valuation multiples, and both were visible in the move.
There is also a timing problem specific to Google's position. The company has spent the better part of two years convincing investors that its AI investments would convert into Cloud growth, and the conversion is now happening - Cloud grew 82% in the latest quarter. But a price cut on a flagship AI product, delivered while the stock is already under pressure, reads to a marginal seller as evidence that monetization will require discounting. The 30% cut is strategically sound for share capture and commercially rational given the cost curve; it is also, in the short run, a reminder that AI revenue may arrive with less gross margin than the bull case assumed.
The strongest version of the bear case runs like this: AI video is following the same path as cloud compute, where performance per dollar doubles on a clockwork cadence and the surplus accrues to the customer rather than the vendor. If every major model lab can generate a consistent 15-second clip, then clip generation becomes an input, not a product - and inputs get priced at marginal cost. Google's $0.07 rate would then be not a weapon but a confession. Under that reading, the winners are the advertisers and creators who buy the seconds, not the labs that sell them, and Alphabet's multiple has room to compress further even as revenue grows.
The counter to the counter is equally concrete. Commoditization happens at the layer where differentiation is impossible, and video is not just generation - it is reference adherence, editing workflow, brand safety, and integration into the tools creators already use. Google Vids and Gemini for business bundle the model into an enterprise environment where switching costs are real. A 15-second clip that keeps a model's face and a garment's drape consistent across 360-degree turns is not a commodity output; it is a workflow asset. The price cut is the hook; the retention happens in the editing loop, the asset library, and the billing relationship.
Cyclical or Structural: This Is a Regime Shift, Not a Promotion
The central judgment of this release is that the price cut is structural, not cyclical. A cyclical price move is a promotion designed to clear inventory or hit a quarterly target, and it reverts. A structural price move reflects a permanent change in the cost curve or the competitive structure, and it does not revert on its own.
Three pieces of evidence point to structural. First, the cost floor has moved: video inference is still roughly 60 times the cost of an image model, but that ratio is falling as models get more efficient and as purpose-built silicon scales. A 30% cut funded by efficiency gains is permanent until a competitor matches it. Second, the competitive structure has changed: OpenAI's exit from subsidized video and the concentration of viable providers at a handful of well-capitalized labs means the market is settling into an oligopoly where price is set by the lowest-cost producer with distribution - which is Google's historical comparative advantage. Third, the buyer behavior is changing: enterprises are not experimenting with one-off clips; they are building catalogs, virtual try-ons, and ad spots that require repeatable, consistent output billed through an API. That is a recurring-revenue relationship, not a trial.
The cyclical counterweight is real but smaller. In the short run, the 15-second ceiling is still a constraint for long-form content, and reference adherence at 360-degree turns remains a demonstrated capability rather than a proven one at broadcast scale. If the quality does not hold up in production, the price cut becomes a discount on a product buyers do not yet trust - a cyclical promotion that fails to convert. That is the risk that keeps this from being a one-way call.
Second-Order Effects: What the Market Is Not Asking
The first-order effect of the announcement is obvious: cheaper video. The second-order effect is what it does to the rest of the AI stack. If a 15-second branded clip costs about a dollar to generate, then the binding constraint on video output shifts from generation cost to distribution and creative direction. Agencies and brands will produce an order of magnitude more variants, and the value will migrate to the layers that select, test, and place those variants - media buying, analytics, and the creative brief itself.
The third-order effect is an expectation gap. The market has priced AI as a margin-expansion story, but the evidence is accumulating that it will arrive as a volume story first. Google Cloud's 82% growth shows the volume is real; the 30% cut shows the margin will be fought over. Investors who bought the multiple-expansion thesis are being asked to underwrite a throughput thesis instead, and that is a different investment case with a different valuation anchor.
What to Watch: The Signals That Will Settle the Debate
Three signals will determine whether this price cut is a share grab or a margin surrender. First, Google's next quarterly disclosure of Cloud growth and any breakout of generative AI contribution - if Cloud growth stays above 50% year over year while the price cut is in effect, the volume story is winning. Second, the pricing actions of the remaining providers: if Luma, Runway, and the Chinese providers match the cut within a quarter, the market is a commodity; if they hold price and compete on quality, differentiation survives. Third, enterprise adoption of the editing workflow - multi-turn conversational editing is the retention hook, and usage growth there would confirm that Google is selling a workflow, not just seconds.
The falsifying signal for the structural thesis is specific: if Google Cloud growth decelerates below 40% year over year in the next two quarters while the price cut remains in place, the volume-offset argument fails and the cut is a margin giveaway, not a moat-builder. Conversely, if two or more premium providers hold their price and grow usage, the differentiation thesis survives.
The Bottom Line
Short term, the stock reaction is the rational read: a price cut on a marquee AI product is margin-negative until proven otherwise, and the shares are already under pressure. Medium term, the volume story in Cloud is real and the price cut is the tool that widens the funnel. Long term, the winner in AI video will not be the lab with the longest clip but the one with the lowest cost per usable second and the stickiest editing workflow - and Google has just moved to own both.
The market sold Alphabet on the day because it heard a discount. The more consequential reading is that Google just set the price floor for an entire category - and dared every other lab to match it.
Explore more exclusive insights at nextfin.ai.
