TechCrunch: ChatGPT’s growth plateaus as Sensor Tower shows Gemini closing the gap on downloads, MAUs, and time spent

## Global trend at a glance TechCrunch, dated December 5, 2025, reports slowing mobile momentum for ChatGPT. Sensor Tower still shows ChatGPT leading on downloads and MAUs. It accounts for roughly 50% of global AI‑chatbot downloads and about 55% of global monthly active users. Google’s Gemini is closing the gap, and its growth rates now outpace ChatGPT on downloads, MAUs, and time spent. OpenAI’s internal “code red” memo underscores urgency on product quality and personalization. The leadership remains intact, but incremental gains are decelerating. Competition is intensifying across categories and regions. Morocco will feel these dynamics in 2026. Through November 2025, ChatGPT’s global MAUs grew about 180% year over year. Gemini’s MAUs rose about 170% in the same period. Both are still expanding at large scale. The near‑term trend tells a different story. From August to November, ChatGPT’s MAUs grew only about 6% to roughly 810 million. Gemini’s MAUs jumped about 30% in that window. Its image model, branded “Nano Banana,” is buoying adoption. Share shifts inside the cohort mirror the change. From May to November, Gemini gained about three percentage points of MAU share. From August to November, ChatGPT’s MAU share slipped by about three points. Downloads show similar pressure. ChatGPT’s global downloads were up about 85% year over year as of November, but the cohort average grew about 110%. Perplexity was up about 215% year over year on downloads, and Gemini was up about 190%. Several rivals posted triple‑digit 2025 growth on broader metrics. Sensor Tower estimates Perplexity at about +370% year over year and Anthropic’s Claude at about +190% year over year. Engagement is tilting as well. Gemini users spent about 11 minutes per day in November, up about 120% since March. ChatGPT’s time in app rose only 6% over that span and fell 10% in November versus July. Sensor Tower adds a structural twist in the United States. Roughly twice as many Android users access Gemini via OS‑level entry points than through its standalone app. OS‑level integration matters most where Android dominates. That includes Morocco. Distribution advantages can compound faster than standalone app installs. Image features and vertical strengths are new growth levers. ## Why this matters for Morocco Morocco’s mobile internet runs heavily on Android devices. OS‑level entry points can tilt adoption toward Gemini if these touchpoints arrive locally. ChatGPT still leads globally and remains strong for text and coding tasks. But multi‑model strategies are becoming the default in competitive teams. Moroccan startups should treat the market as multi‑model by design. They can calibrate Gemini for image workflows and discovery. They can keep ChatGPT for general reasoning, code, and content. They can add Claude or Perplexity for research and analysis. ## Policy and public sector momentum Government bodies continue pushing digital transformation. The Agency for Digital Development has promoted e‑government services and modernization. Sector ministries are digitizing licensing, payments, and citizen interfaces. AI guidelines are under discussion as teams pilot automation across functions. Public procurement can catalyze local language innovation. Tools must handle Moroccan Arabic (Darija), French, and Amazigh. That requires fine‑tuning prompts, lexicons, and datasets. It also demands consistent guardrails and quality monitoring. ## Startups and builders Morocco’s startup scene is maturing across Casablanca, Rabat, Tangier, and Marrakech. Incubators and innovation hubs are supporting AI experimentation. Seed funding is growing, with more corporate partnerships in fintech, telecom, and retail. Universities are expanding AI coursework and applied labs. Founders are reading the same global signals. Gemini’s image strength is attractive for consumer experiences. ChatGPT remains a workhorse for developer co‑piloting and support bots. Teams are mixing models pragmatically to improve outcomes and costs. ## Practical AI uses already delivering value ### Agriculture and water AI can forecast yields, detect pests, and optimize irrigation schedules from satellite and sensor data. Farmers can reduce water stress with predictive alerts. Cooperatives can plan seasonal logistics with better accuracy. These gains matter in drought‑sensitive regions. ### Tourism and hospitality Multilingual chatbots enhance booking support in Arabic, French, and English. Generative translation reduces friction for foreign visitors. Image search and itinerary generation can highlight attractions dynamically. Hotels can route requests with AI triage to cut wait times. ### Retail and e‑commerce Image models can auto‑tag products and improve visual search. Merchants can generate descriptions and size guides from photos. Support bots can handle returns and exchange flows with clearer policies. Dynamic pricing engines can respond to demand and inventory signals. ### Financial services Banks and fintechs can deploy fraud models for card and account monitoring. Onboarding assistants can pre‑check documents and KYC fields. Contact centers can use AI for summarization and next‑best action. Credit scoring teams can test explainable models with local variables. ### Public services and municipalities Document classification speeds up case routing and archiving. Citizen assistants handle appointments and FAQs across channels. Computer vision can track traffic flows and improve safety planning. Waste operations can use recognition to guide sorting and recycling. ### Education and training Teachers can build lesson plans with generative templates. Students can access tutoring and exam preparation tools. Language learning improves with conversational AI in Darija and French. Universities can integrate AI sandboxes for capstone projects. ## Choosing between ChatGPT, Gemini, Claude, and others Start with the job to be done. Define tasks, latency needs, and content sensitivity. Plan for a multi‑model architecture with clean abstractions. Use API gateways to swap models without touching your core logic. Evaluate six dimensions consistently. Cost per request and expected monthly spend. Text reasoning quality on your domain data. Image generation and editing for consumer experiences. Language coverage for Darija, French, and English. Privacy controls and regional data handling. OS‑level integrations and edge features on Android. Test care paths in real workflows. Pilot live with a small segment of customers or staff. Log prompts, outputs, and failure modes. Iterate weekly with tight feedback loops. ## Measurement and engagement Adopt the same metrics highlighted by Sensor Tower. Track monthly active users per initiative. Track time in app or time on task for each AI surface. Watch acquisition sources for OS‑level entry points on Android. Segment by language and device class. Monitor stickiness and cohort retention. Judge success by business outcomes, not only model scores. For example, reduced support wait times or higher conversion rates. ## Governance and risk Set clear policies for data retention and human review. Use content filters to block sensitive outputs. Build audit trails for decisions affecting customers. Train teams on responsible use, bias, and privacy. Negotiate vendor contracts with portability in mind. Avoid committing all workflows to a single model. Maintain fallbacks so outages do not halt operations. Review changes when models update and retrain. ## Distribution and product design in Morocco Android is widespread, so distribution should favor OS‑level shortcuts and intents. Deep‑link from notifications and system search into your AI surfaces. Keep image shortcuts prominent in commerce and travel apps. Optimize multilingual flows for Darija and French first. Use lightweight prompts for low bandwidth contexts. Cache frequently used responses with local storage when possible. Keep accessibility front‑and‑center in government and service apps. Small design choices compound into daily time‑spent wins. ## 2026 outlook ChatGPT remains the most widely used mobile AI chatbot globally. Leadership is intact, but incremental growth has slowed versus peers. Gemini’s trajectory on acquisition and stickiness is favorable near term. OS‑level integration and differentiated image features are clear levers. TechCrunch frames the picture as a competitive, multi‑model market. The cohort’s faster growers are chipping away at share. OpenAI could re‑accelerate if upcoming releases deliver outsized utility. Moroccan teams should stay agile and build with model diversity from day one. ## Key takeaways - ChatGPT leads on mobile but shows slowing incremental growth. - Gemini is gaining share, helped by Android entry points and image features. - Morocco should adopt a multi‑model approach across sectors. - Prioritize multilingual support for Darija, French, and English. - Measure MAUs, time spent, and business outcomes to guide investment. - Build governance, portability, and fallbacks into every AI deployment. - Expect a more competitive, multi‑model market in 2026.