meta's chief ai officer alexandr wang is closing out his first year – and it's not going well

why he's being criticised:

• leadership style called chaotic and impatient
• accused of hyping small, step-by-step improvements as if they were major breakthroughs
• talent leaks – a former apple exec left for openai after just seven months
• an early attempt to build a fresh training codebase stumbled; muse spark ended up leaning on existing llama 4 code and data
• his tenure overlapped with layoffs and a controversial employee-tracking plan meta had to partly walk back
• even wang and zuck reportedly set low internal expectations for muse spark
• some staff doubt meta can reach the frontier under him at all

wang's plans:

• tbd lab – wang's secretive 100-person elite research unit inside meta – plans to release more models in the coming months to narrow the gap with openai, google and anthropic
• focus next on coding (muse spark lags here – staff testing it for dev work still reach for claude)
• push into agentic task execution + advanced multimodal, including video gen
• shift toward proprietary models over meta's open-source tradition
• bigger pitch: wang's "personal superintelligence" vision

this reads like a roadmap toward a commodity model. muse spark benchmarks around deepseek v4 pro level per artificial analysis, but deepseek undercuts everyone on price. why would anyone pick meta's model over the cheaper one that does the same thing?

the only real path left is niching down – absent a genuine breakthrough, being merely good (but not better than rivals, and not cheaper) is a rough spot to be in

source: bitget