The end of technological autarky: Meta relies on Midjourney, Apple looks to Gemini to relaunch Siri

The monopoly of technological pride is over. After years in which Apple and Meta - as well as the other Silicon Valley giants - have claimed self-sufficiency in their own research labs - relying on Cupertino's vertical integration and internal projects such as the FAIR lab and Menlo Park's LLaMA models - today both giants are choosing the path of alliances. Meta has announced a licensing agreement with Midjourney to integrate its technology into future products; Apple has started preliminary talks with Google for a possible integration of Gemini into a redesigned Siri. Not even big tech can afford to chase innovation alone.

Meta's decision to enter into an agreement with Midjourney marks a change from the tradition of in-house development. Midjourney, which was founded in 2021 and went public in 2022,has quickly established itself as a leader in generative artificial intelligence (AI) for the production of high-impact images. Its models, disseminated through a global community, have shaped an instantly recognisable style, somewhere between digital art and photographic realism.

Mark Zuckerberg's giant tried to chase with proprietary systems such as Imagine (image generation) and the experimental Movie Gen model/stack for video, but these struggled to keep up with rivals such as OpenAI's Sora or Google's Veo. For a company that thrives on visual content - from social sharing to advertisements - the gap risked becoming an unsustainable competitive factor. Hence the decision to open the door to an external technology, destined to flow into the future creative tools of the Meta ecosystem (on its main social platforms).

The new Chief AI Officer, Alexandr Wang, spoke of an all-of-the-above strategy, combining in-house research, computing infrastructure and targeted alliances. It is a cultural shift that marks the end of an idea (and ideal) of total self-sufficiency. However, legal questions remain open because Midjourney has been repeatedly accused of training its models on datasets containing copyrighted works. Meta knows that the risks are real, but it has chosen the quickest route to become competitive again in the field of digital aesthetics.

According to Grand View Research, the global generative artificial intelligence market is estimated at $16.87 billion in 2024 and could exceed $100 billion by 2030. For Meta, falling behind does not only mean losing prestige, but compromising its main source of revenue: in Q2 2025, the group recorded $46.56 billion in advertising revenue out of a total of $47.52 billion, or about 98 % of revenues.