Karen Hao has spent seven years overlaying synthetic intelligence improvement, culminating in a conclusion that challenges typical understanding of the expertise sector. OpenAI and comparable organizations must be analyzed as empires somewhat than companies, in response to the investigative journalist’s analysis printed in “Empire of AI.”

“These corporations I actually assume must be considered new types of empire,” Hao said throughout a September 2025 interview on The Room The place It Occurred podcast. “They are not simply working within the enterprise sphere. They’re additionally working within the political sphere.”

The empire framework emerged from Hao’s 2019 embedding inside OpenAI workplaces when she was reporting for MIT Know-how Assessment. That three-day investigation revealed a company functioning past typical company parameters whereas deploying billions of {dollars} based mostly on what she characterizes as quasi-religious ideology.

Subscribe PPC Land e-newsletter ✉️ for comparable tales like this one. Obtain the information day by day in your inbox. Freed from adverts. 10 USD per 12 months.


Subscribe

Imperial traits outline AI improvement

Hao identifies particular attributes that distinguish AI corporations from conventional companies. These organizations accrue “an unlimited quantity of social, political, financial capital” whereas buying land, power, water, and uncooked sources to construct expertise at scales akin to historic empires.

“One of many driving motivators behind what OpenAI does is a form of this quasi non secular ideology that synthetic normal intelligence is true on the horizon and it is one way or the other going to deliver profound profit to humanity,” Hao defined. “And that is additionally very imperial in nature.”

The parallel to historic imperialism extends to ideological justification. In line with Hao, empires of outdated have been “pushed by a non secular ideology of we have to deliver progress and modernity to all of humanity. And that is why we have to interact in grabbing all this land, grabbing all these uncooked sources, exploiting labor, constructing these applied sciences that finally fortify the empire and will probably find yourself exacerbating world inequalities.”

OpenAI’s mission operates identically, she argues. The ideological element means totally different people inside the group interpret their work by means of private worth techniques, creating fragmented understanding of precise aims and strategies.

“Each individual form of has their very own private worth system that they are utilizing as a prism by means of which to see OpenAI’s mission and find out how to obtain it,” Hao said.

Quasi-religious mission drives useful resource accumulation

The idea construction inside AI corporations goes past typical enterprise motivation. Hao describes workers who “assume that they’re making a a digital god or defending us from a digital demon. and that that is their calling, that they have been delivered to this earth to usher in a brand new type of intelligence that would be the subsequent stage of the human species.”

This ideological basis allows unprecedented useful resource consumption. The AI business requires including power equal to 0.5-1.2 instances United Kingdom annual consumption to world grids inside 5 years, in response to Hao’s analysis. Most extra capability will derive from coal and fuel crops beforehand scheduled for retirement.

“The size has simply fully exploded to an unfathomable diploma,” Hao said. “And and generally I really feel just like the the dimensions and scale and numbers that these tech CEOs throw out at individuals is purposely so giant and extraordinary that the typical individual on the earth has by no means encountered one thing of that magnitude and subsequently it simply turns into incomprehensible.”

Meta CEO Mark Zuckerberg exemplifies this sample. He said the corporate’s AI supercomputers would approximate Manhattan’s measurement and power consumption. “That’s one thing that’s unprecedented and that’s single-handedly at the moment reversing a unprecedented quantity of local weather progress that we made within the final decade,” in response to Hao.

Purchase adverts on PPC Land. PPC Land has customary and native advert codecs by way of main DSPs and advert platforms like Google Advertisements. Through an public sale CPM, you’ll be able to attain business professionals.


Learn more

Strategic authorities alignment mirrors historic patterns

Hao attracts express comparisons to the British East India Firm’s relationship with the British Crown. “You had a company empire and a nation state empire that have been strategically aligned and benefiting from each other in a manner that basically finally led to a variety of exploitation and hurt all over the world,” she defined.

The Trump administration’s strategy to AI corporations demonstrates comparable dynamics. “The US authorities has made underneath the Trump administration very clear that they need to rebuild the American empire and broaden and strengthen it. And so they’re they see these American tech corporations as instruments for doing in order shortly as potential,” Hao said.

Authorities officers are “making an attempt to deploy OpenAI’s applied sciences all all over the world and so they’re making an attempt to assist OpenAI construct extra knowledge facilities to facilitate that,” in response to her evaluation. Former Dutch parliamentarian Marietje Schaake famous in a Monetary Instances op-ed the strategic benefit of putting in American {hardware} and software program globally with potential distant management capabilities.

“I believe that’s a part of the rationale why the US authorities is is unwilling to do something to test these corporations as a result of it it believes that it could possibly preserve management over these organizations and proceed to make use of them as fingers to perpetuate like their very own political will,” Hao said.

Democracy faces erosion from company energy

Hao’s biggest concern facilities on democratic governance deterioration. “These corporations, OpenAI, however the entire different Silicon Valley giants and tech and AI giants, they’ve grown so highly effective that they are surely starting to fully warp and deform democracy,” she said.

The businesses can “purchase land, create digital currencies, um, take individuals’s power and water, hike up their utility payments, distort individuals’s instructional and financial alternatives simply with the wave of a hand based mostly on what they resolve to deploy of their merchandise someday to the following based mostly both on enterprise or ideological grounds.”

The Trump administration alliance with Silicon Valley eliminates what was “arguably initially the one entity that h that was extra highly effective than Silicon Valley and will truly regulate these corporations,” in response to Hao. “And now underneath the Trump administration, we see completely no curiosity from the US authorities in in taking part in that counterveailing pressure.”

This trajectory threatens elementary democratic ideas. “If we proceed to be on this trajectory the place these corporations get increasingly more and increasingly more highly effective and increasingly more and extra rich the place they will simply purchase up no matter they need and and knock over no matter no matter obstacles of their manner that we’re going to lose the flexibility to collectively self-determine our futures and that is when democracy dies,” Hao said.

Techno-authoritarian governance replaces democratic enter

The expertise business operates essentially in another way from democratic techniques, in response to Hao’s evaluation. “There also needs to be elected representatives that characterize all of us. And there must be a democratic type of governance round these extraordinarily consequential points and the way they need to be resolved and the way we deal with them.”

As an alternative, the business features as techno-authoritarianism. “Open AI does not function that manner. Meta does not function that manner. Google, Microsoft, Apple, they’re they function in a way more techno athoritarian manner the place they dictate to billions of customers or in OpenAI’s case, lots of of hundreds of thousands of customers precisely what their techn like these these profoundly influential algorithms are going to do and combine into their lives and and form the course of their healthcare, their financial alternatives, their instructional alternatives.”

The implications prolong past particular person platforms. “We we have to have a say in this stuff which might be molding who we change into as a society,” Hao said.

Nonprofit construction served strategic functions

Hao characterizes OpenAI’s nonprofit origins as probably strategic from inception, addressing sequential bottlenecks somewhat than representing real mission dedication. When based in late 2015, Google dominated AI analysis expertise. OpenAI “could not compete on compensation. And they also competed on a way of mission and the nonprofit was a extremely handy sign of how dedicated OpenAI was to that mission.”

Chief Scientist Ilya Sutskever “explicitly got here to OpenAI due to that mission. He took an enormous pay lower to come back to OpenAI,” in response to Hao. As soon as core groups assembled, useful resource necessities shifted. “The bottleneck turned about capital as a result of now it had all of the it had a core group of actually sensible individuals and it simply wanted to construct.”

The for-profit transition addressed this new constraint. “That is once they thought nonprofit does not lower it anymore. We’ll put a for-profit arm inside this nonprofit and use it as a car to lift billions of {dollars},” Hao defined.

“Should you have a look at that, you are like, nicely, that form of looks like it was the technique all alongside, you already know, such as you ch you first design the group to deal with one bottleneck after which as soon as you’ve got solved that bottleneck, you morph it to deal with the following bottleneck,” she said.

Present inequality indicators future trajectory

Observable outcomes contradict guarantees of broadly distributed AI advantages. “The AI increase has created extra billionaires at a sooner price than ever earlier than and has left new faculty graduates with an incapacity to search out financial alternatives,” Hao said.

“What I see at the moment is that’s creating extra inequality, extra divisiveness and that must be the strongest sign for what would come subsequent if we do not take our future in our personal fingers and alter the trajectory of AI improvement,” she defined.

Sam Altman has claimed AI productiveness positive factors may allow $13,500 annual funds to each grownup. Hao responded: “Nicely, what I do know is what we’re seeing now, which is that the AI increase has created extra billionaires at a sooner price than ever earlier than.”

Collective resistance stays potential

Regardless of concentrated energy, Hao emphasizes democratic affect potential. AI corporations require elements collectively owned by society, together with knowledge from artists, writers, and social media customers. “There are that is the information of artists and writers and other people posting on social media, individuals posting their children photographs on platforms like Flickr,” she said.

Copyright lawsuits characterize profitable resistance. “We’re seeing important push again from a few of these teams saying no, we’ll sue you as a result of we won’t settle for that you simply’re simply going to take our IP that we’ve we’ve copyright protections on this work. And that’s them reclaiming possession over a key ingredient that this firm wants entry to.”

Neighborhood protests have delayed knowledge middle tasks globally. “We’re seeing lots of of protests breaking out all over the world with individuals pushing again in opposition to knowledge middle improvement as a result of they’re saying we do not need to host an information middle in our neighborhood. It is climbing up our utility payments. It is creating all these different environmental impacts.”

Academic debates about AI integration display one other resistance level. “We’re seeing college students and academics have big discussions about whether or not or not we truly need AI in training, whether or not AI is definitely enabling higher training or it is simply degrading individuals’s important pondering abilities.”

OpenAI responded to criticism about ChatGPT eroding scholar analytical talents by including research modes. “So like all of those I I usually have a look at like AI improvement. There is a provide chain identical to every other product,” Hao said.

Know-how improvement requires democratic participation

Hao’s elementary message challenges assumptions about technological inevitability. “I really need individuals to know that expertise improvement is just not inevitable and that it’s also not simply the purview of those that work in tech corporations,” she said.

“Truly, each single individual, no matter the place you sit in society, can have a say in expertise improvement as a result of when you concentrate on how these corporations develop these AI fashions, they really want as a result of it is so useful resource intensive, they want entry to a complete host of various elements which might be owned collectively owned by totally different teams in society.”

The framework positions varied resistance factors as democratic contestation websites. “If we take into account these as totally different websites of democratic contestation, then we’ll if and we’ve totally different actions, totally different teams which might be pushing to really assert what they need out of the expertise, how they need the expertise developed, all in live performance with each other. Firms have to reply,” Hao defined.

She explicitly rejects counting on authorities intervention. “I do not assume individuals ought to wait for presidency. The attractive factor about democracy is that even when there’s an absence of management on the high, there’s management from the underside.”

Scale distinguishes present second from precedents

The useful resource necessities separate AI improvement from earlier expertise sectors by orders of magnitude. Meta gathered roughly 4 billion person accounts of knowledge in the course of the social media period, but executives “have been speaking about along with that knowledge, what about if we purchased out Simon and Schustster for all of their books in order that we may prepare a generative AI system?”

The corporate finally acquired books by means of unauthorized channels. “Ultimately, they did not purchase Simon and Schustster. They only tormented all of the books from the darkish internet and used it to coach their AI fashions as a substitute,” Hao said.

“Even even in comparison with social media which already we are able to argue was problematic in its scale. AI is manner worse,” she defined. The unprecedented magnitude allows operations that slip “underneath the radar” as a result of common individuals can not comprehend the size concerned.

Subscribe PPC Land e-newsletter ✉️ for comparable tales like this one. Obtain the information day by day in your inbox. Freed from adverts. 10 USD per 12 months.


Subscribe

Timeline

Subscribe PPC Land e-newsletter ✉️ for comparable tales like this one. Obtain the information day by day in your inbox. Freed from adverts. 10 USD per 12 months.


Subscribe

Abstract

Who: Investigative journalist Karen Hao performed seven years of AI business analysis, together with 2019 embedding inside OpenAI workplaces whereas at MIT Know-how Assessment. The investigation examined CEO Sam Altman, Chief Know-how Officer Greg Brockman, Chief Scientist Ilya Sutskever, and broader organizational dynamics throughout the AI sector.

What: Hao argues AI corporations together with OpenAI operate as fashionable empires somewhat than typical companies, working throughout political, social, and financial spheres whereas accumulating sources at imperial scales. The quasi-religious ideology driving synthetic normal intelligence improvement mirrors historic imperial justifications for useful resource extraction and labor exploitation. Strategic evolution from nonprofit to for-profit construction probably served recruitment and capital acquisition aims whereas sustaining public relations advantages from unique mission-driven branding.

When: Preliminary investigation occurred in 2019 throughout OpenAI’s nonprofit part, earlier than November 2022 ChatGPT launch and subsequent industrial growth. Analysis spanned seven years by means of September 2025 e-book publication, documenting the group’s transformation and broader business patterns throughout AI’s mainstream adoption interval.

The place: Investigation centered on OpenAI’s San Francisco workplaces with evaluation extending to world AI improvement infrastructure together with worldwide knowledge facilities, power techniques, and useful resource extraction operations. The empire framework examines operations spanning political relationships with the Trump administration and deployment methods throughout worldwide markets.

Why: Hao initiated analysis to confirm organizations pushing AI frontiers and look at alignment between said missions and precise practices. The empire idea issues for advertising professionals as AI reshapes advertising strategies, platform commerce capabilities, and campaign management infrastructure. Understanding imperial dynamics and ideological motivations helps entrepreneurs assess democratic implications, consider platform dependencies, and acknowledge collective resistance potential by means of knowledge possession, neighborhood organizing, and academic debates shaping expertise improvement trajectories.


Source link