Meta Chief Executive Officer Mark Zuckerberg has acknowledged that the company’s artificial intelligence agents are not advancing as quickly as leadership had anticipated. The admission comes despite sweeping organisational changes, billions of dollars in AI spending and one of the technology industry’s most aggressive pushes toward artificial intelligence.
Meta CEO Mark Zuckerberg has admitted that the company’s AI agent technology is developing more slowly than expected, offering a rare glimpse into the internal challenges facing one of the world’s biggest artificial intelligence initiatives.
During an internal town hall meeting, Zuckerberg told employees that AI agents had not “accelerated in the way” company executives had hoped, according to reports first published by Reuters and highlighted by TechCrunch. The comments come as Meta continues investing heavily in AI infrastructure, talent and products in an effort to establish itself as a leader in the next generation of artificial intelligence.
The acknowledgement marks a notable shift in tone from a company that has spent the past year presenting artificial intelligence as its central growth strategy.
Meta has reorganised thousands of employees, invested billions of dollars in AI infrastructure and recruited leading researchers from across the industry. Those efforts were intended to accelerate the development of AI-powered systems capable of completing complex tasks on behalf of users.
According to the report, Zuckerberg also acknowledged that the company’s broader restructuring had not gone as smoothly as executives had planned.
«”The systems known as AI agents had not progressed as quickly as I had expected,” Zuckerberg told staff during the meeting, while also admitting the reorganisation was not as “clean” as it could have been.»
The restructuring included laying off about 10% of Meta’s global workforce and reassigning roughly 7,000 employees to AI-focused teams earlier this year. While those changes were designed to sharpen the company’s focus on artificial intelligence, they also prompted internal concerns about morale and organisational stability.
Meta has been pouring unprecedented resources into artificial intelligence. The company is expected to spend around $145 billion on AI infrastructure this year, funding advanced data centres, specialised chips and large-scale computing resources needed to train increasingly sophisticated AI models. Those investments are also supporting Meta’s long-term ambition to develop AI assistants capable of performing real-world tasks with minimal human input.
AI agents have become one of the industry’s biggest areas of focus. Unlike conventional chatbots that primarily answer questions, AI agents are designed to perform actions on behalf of users. They can schedule meetings, book travel, manage workflows, conduct research and interact with other software autonomously.
See Also: Palantir CEO Alex Karp warns AI’s biggest battle won’t be better models, but who owns the data
Technology companies including OpenAI, Google, Microsoft, Anthropic and Amazon are all investing aggressively in this emerging category. Meta has positioned AI agents as a key part of its long-term strategy, integrating them into products across Facebook, Instagram, WhatsApp and its growing ecosystem of AI-powered devices.
Despite slower-than-expected progress, Zuckerberg reportedly told employees he still expects meaningful improvements over the next three to six months as the company’s investments begin producing stronger results. The internal discussion also addressed another issue that recently generated employee concern.
Meta’s Chief Technology Officer Andrew Bosworth said the company had paused a controversial programme that monitored employee mouse movements and computer activity to help train AI systems. Following internal criticism, the initiative may return as an opt-in programme rather than being mandatory. Bosworth acknowledged that the rollout negatively affected employee morale, although he noted the project produced valuable technical insights.
The challenges highlight the complexity of developing advanced artificial intelligence at scale. Building frontier AI systems requires enormous computing resources, specialised engineering talent and continuous experimentation. Even with significant financial investment, breakthroughs often take longer than anticipated.
Meta’s experience also demonstrates that organisational changes alone cannot guarantee faster technological progress. As companies across the industry compete to develop increasingly capable AI systems, attracting top researchers, expanding infrastructure and restructuring teams are only part of the equation.
Execution remains equally important. Industry analysts believe Zuckerberg’s comments provide an unusually candid assessment of the realities facing AI development.
Public announcements often focus on new models and ambitious product roadmaps, while the technical and organisational setbacks behind those projects receive far less attention. Meta’s willingness to acknowledge slower progress suggests the company recognises that expectations surrounding AI have become exceptionally high.
Investors, employees and customers are watching closely to see whether massive spending on artificial intelligence will translate into practical products capable of generating meaningful returns.
For now, Zuckerberg remains committed to the company’s long-term AI strategy. His remarks suggest Meta still believes AI agents will become a central part of how people interact with technology, even if reaching that goal is proving more difficult than originally expected.
As competition intensifies across the AI industry, the companies that succeed may not necessarily be those spending the most money. They are likely to be the ones that can consistently turn ambitious research into reliable products that deliver measurable value for businesses and consumers alike.
