AI poses a new antitrust problem - FT中文网
登录×
电子邮件/用户名
密码
记住我
请输入邮箱和密码进行绑定操作:
请输入手机号码,通过短信验证(目前仅支持中国大陆地区的手机号):
请您阅读我们的用户注册协议隐私权保护政策,点击下方按钮即视为您接受。
人工智能

AI poses a new antitrust problem

Washington cheered on the consolidation of the social web in the 2010s — it must not make the same mistake today
00:00

{"text":[[{"start":null,"text":"

By taking controlling stakes in AI companies, hyperscalers are ensuring that the industry evolves in a way that serves their infrastructure
"}],[{"start":6.69,"text":"The writer is former associate director for litigation in the FTC’s Bureau of Competition and co-founder of Simonsen Sussman"}],[{"start":15.830000000000002,"text":"I joined the Federal Trade Commission in the summer of 2021, just weeks after a federal judge dismissed the agency’s antitrust complaint against Facebook. It was a frantic, all-hands-on-deck moment. I was thrown into a room with lawyers and economists and tasked with keeping the case alive."}],[{"start":36.91,"text":"We rewrote the complaint, which challenged Facebook’s acquisitions of Instagram and WhatsApp in 2012 and 2014 respectively. Even then we were well aware of how late it was. If the FTC eventually won at trial (which, spoiler, it did not), the appeals process and complexity of unwinding these companies meant that a remedy might not arrive until 2029 — 17 years after the first acquisition."}],[{"start":67.21,"text":"Last month, Judge James Boasberg ruled that Meta was not a monopoly. The decision highlighted the difficulty that regulators around the world have in challenging acquisitions long after their completion."}],[{"start":81.6,"text":"Yet when Facebook made these purchases, regulators offered little pushback. This reflected a broader pathology at the time. In the wake of the 2008 financial crisis, the American economy seemed precarious. Against that backdrop, Silicon Valley offered a glimmer of hope. Companies like Facebook and Google were viewed as national champions. CEOs were paraded through the White House, treated as dignitaries of a new economic order."}],[{"start":113,"text":"We are now watching history repeat itself. Just as the government cheered on the consolidation of the social web in the 2010s, Washington is offering similar support today for the “hyperscalers” — Microsoft, Amazon and Google — to position themselves as guardians of artificial intelligence. Despite tough rhetoric, the administration is not pushing back on AI deals. We saw this recently when the Department of Justice cleared Google’s acquisition of cloud security company Wiz, waving through further consolidation in the cloud security market. Similarly, the regulatory response to the $38bn cloud deal between Amazon and OpenAI — an arrangement that binds the most prominent AI start-up to a dominant cloud incumbent — has been muted."}],[{"start":166.03,"text":"Such deals are fraught with conflict of interest. By taking controlling stakes in AI companies, hyperscalers are ensuring that the industry evolves in a way that serves their infrastructure and increases the moats around their lines of business. They have every incentive to direct innovation towards massive, data-hungry, compute-inefficient models rather than efficient, decentralised systems."}],[{"start":192.73,"text":"Yet instead of scrutiny, we are seeing a pilgrimage of AI CEOs to the White House. Proposals for a moratorium on state AI laws have been made. This is the language of antitrust amnesty. It suggests that, once again, economic uncertainty is leading Washington to trade competition for the pursuit of international corporate leadership."}],[{"start":216.26999999999998,"text":"The contrast between the current administration and the previous one is stark. When Nvidia attempted to acquire Arm in 2020 — a deal that would have locked down the fundamental architecture of chips used for AI — the FTC sued to block it and the deal was abandoned. Because Arm remained independent, Google is now able to compete with Nvidia in the production of AI chips, using Arm’s technology to build its own processors.  "}],[{"start":247.74999999999997,"text":"Regulators waited too long to address the danger that Facebook’s acquisitions posed to competition. This should be a wake-up call for the AI era. If we allow hyperscalers to devour the AI ecosystem under the guise of national competition, we will be left with a tech sector that is more centralised and more extractive than the one we have today. By the time it is a problem, it may once again be too late to change."}],[{"start":283.82,"text":""}]],"url":"https://audio.ftcn.net.cn/album/a_1765248116_1543.mp3"}

版权声明:本文版权归FT中文网所有,未经允许任何单位或个人不得转载,复制或以任何其他方式使用本文全部或部分,侵权必究。

政治脆弱性给市场造成的代价

自各国央行在2022年退出量化宽松以来,主权债务市场对政治脆弱性的敏感度大幅上升。欧洲的政治风险正日益成为核心投资组合构建的一部分。
1小时前

AI热潮中经济学家能否规避利益冲突?

凯恩斯:研究AI、与AI实验室合作成了经济学家的新热门,但经济学家能否保证自己的研究不被扭曲?

Lex专栏:今日运动鞋价格战,明日轮到AI巨头

恶性价格竞争或许能让消费者受益,但企业利润率可能在这一过程中崩塌。

富豪疯狂囤金之年

富豪们对黄金的信念,如同这种金属一样未曾褪色。

特朗普令冰岛加入欧盟的争论重新升温

冰岛人将于8月29日投票,决定该国是否应再次尝试加入这一集团。

为什么泰国的利率处于全球最低水平之列?

泰国尚未实现富裕便已步入老龄化,而高负债水平抑制了本可提振经济的消费支出。
设置字号×
最小
较小
默认
较大
最大
分享×