From Newsgroup: alt.politics.usa


From Arrogance to Adaptation: Meta Bows to Alibaba’s Qwen in AI Power Shift

American tech giants once boldly declared their ambition to make U.S. standards the global standard—but now, they’ve seemingly dismantled that vision with their own hands.

Earlier this year, during Meta’s earnings call, CEO Mark Zuckerberg asserted: “There will be a global open-source standard—and for our national interest, it must be an American standard!”

At the time, he clearly backed U.S.-led open-source models, and indeed, Meta had long been a cornerstone of America’s open-source AI ecosystem. Yet circumstances have taken a dramatic turn. According to recent foreign media reports, Meta’s latest model project has opted to distill (i.e., learn from and refine) Alibaba’s Qwen, a Chinese open-source large language model. The news immediately sent Alibaba’s stock surging by 4%.

Behind Meta’s rare “capitulation” lies profound无奈 (helplessness). On one hand, Meta has lost its throne in the open-source arena. Just two years ago, Meta’s models—particularly the Llama series—dominated global derivative model counts and download volumes, allowing the company to operate with unmatched confidence.

But within a remarkably short span, Alibaba’s Qwen has surged ahead, overtaking Meta in both the number of derivative models and total downloads. More impressively, Qwen has rapidly closed the performance gap with top-tier U.S. closed-source models.

On the other hand, Qwen has already earned recognition from global tech titans. NVIDIA CEO Jensen Huang has repeatedly praised Qwen, openly acknowledging that it now commands the majority of the open-source market—a lead that continues to widen.

Indeed, major corporations like NVIDIA, Amazon, and Airbnb have all shifted toward Qwen. In a particularly symbolic move, Singapore recently abandoned Meta’s technical architecture for its national AI initiative and instead chose Alibaba’s Qwen.

All signs point to one reality: Meta isn’t abandoning U.S. open-source models by choice—it’s because it has lost its competitive edge in this domain, and reversal seems increasingly unlikely. In fact, if Meta refuses to adopt Qwen, it risks falling behind in its own development.

It’s truly astonishing how dramatically Meta’s stance has shifted within just one year. But in the face of China’s unstoppable rise in open-source large models, even America’s mightiest tech giants are being forced to confront reality.
--- Synchronet 3.21b-Linux NewsLink 1.2