Over four days in Washington, 500 CEOs, 25 G20 finance ministers, a quarter of the US Senate, and leaders from 90+ countries gave a nearly unanimous answer to the same question: what economy are we building toward?

It's AI. And the race to build it carries the emotional register of the one our parents and grandparents lived through; the internet, the space race, the A-bomb.

Senator Mark Kelly (D-AZ), speaking in the calm tone of a man who has actually flown to space, sketched a decade-long horizon: a US base on the moon, a working lunar economy, and a new one forming on Mars.

His vision raises a bigger question, and a more urgent one: what's already being built right now, whether we're ready for it or not?

What are we building towards?

Senator Kelly took the stage at Semafor's 2026 World Economy Summit and sketched a picture of humanity on the horizon. This is the Manhattan Project of our lifetimes; it just happens to be playing out in boardrooms and hyperscaler purchase orders rather than in deserts. Every fireside chat and panel discussed how much power and how many people we need to get there.

We have not seen another moment in modern business history when every CEO, in every category, has been this synchronized on the same inflection point, simultaneously.

This is AI’s halo effect in action. 

“But I think eventually, if we try to build an economy on the surface of the moon, after we get on to these later Artemis missions, over time, we can transition to that more commercial model, and then NASA can go off and maybe send people to Mars.”
— Senator Mark Kelly, D-AZ

The ‘halo effect’ in brand psychology describes how one dominant trait colors the perception of everything around it. Originally documented by American psychologist Edward Thorndike, it describes a cognitive bias where an overall impression of a person influences how we evaluate their traits. The concept matters as a diagnostic. When one force becomes this gravitationally dominant, it stops being a category story and becomes a civilizational one.

The forum made a compelling case that AI is doing this very thing in the global economy. And the spotlight was on the US, which needs the power grids, transportation networks, labor markets, and capital flows to keep pace. 

Penny Pritzker, former US Secretary of Commerce and co-host of Semafor's CEO Signal show, offered the cleanest framework of the week by grouping AI's effects into three Ps: Power, Productivity, and People. 

• “Power” is the geopolitical contest. Pritzker warned the US is only "months ahead" of China.
• “Productivity” is the diffusion problem. Several sessions challenged whether organizations have the capacity to effectively integrate AI into their workflows and upskill employees effectively.
• “People” represented the workforce transition. If mishandled, it could dwarf the China Shock.

You cannot optimize for one ring of the halo while ignoring the others. A brand investing in AI-powered personalization while ignoring the autonomy and labor transitions reshaping its operations is solving for the wrong variable. It's a compound story worth exploring, and we break it all down in this Member Brief, exclusively for Future Commerce Plus members. 

The Infrastructure Halo: Power and Energy 

Four years ago, nearly all AI computing power was used for training. As LLMs ingested information and “learned,” their full capabilities were hidden from consumers. Today, two-thirds of AI compute is inference, according to Steve Jang, Founder and Managing Partner of Kindred Ventures. Real people and businesses are using these models to produce outcomes. The curve flipped without most of the economy noticing or adapting. 

“AI compute is the new strategic resource of the 21st century,” Jang said. “If the 20th century was oil, we are now dependent upon AI compute.”

Joseph Dominguez, CEO of Constellation Energy, expanded upon how demand from data centers is creating the need for greater power capacity. When ChatGPT first launched in 2022, the company’s two data centers were providing 1-2M megawatt-hours; today, they provide nearly 40M.

“If NIMBYism becomes the reason we lose the AI race, it will lead to a whole lot of trouble in this country.”
— Joseph Dominguez, CEO of Constellation Energy

US infrastructure isn’t designed to keep up with demand. A power plant still takes five to seven years to build. Pedro Pizarro of Edison International called for a "shot clock" on the appeals process. He cited one transmission project that took 17 years to obtain its final permit, and another that took 20 years. The backlog compounds at every layer of the data supply chain: Kate Johnson of Lumen noted that permitting for networking projects alone can take up to 18 months, meaning new fiber miles take nearly three years to install.

Local legislation is adding friction to the federal system. Some states, like Maine, are banning data center projects outright. When asked whether other states would follow, Dominguez was direct: “We can't do this on a state-by-state basis. We have to have a national policy. If NIMBYism becomes the reason we lose the AI race, it will lead to a whole lot of trouble in this country.”

Liz Claman of Fox Business added to the discourse: “And then there’s BANANA, build absolutely nothing anywhere, near anyone.”

Howard Lutnick, the sitting US Secretary of Commerce, put a dollar figure behind the administration’s investments and goals for the future. The plan is to put $18T toward stranded power infrastructure, the dormant grid capacity left behind by factories offshored during the last wave of industrial policy. The goal is to reclaim the power, rebuild the fabs, and resurrect the towns, which will trickle down to create a spike in real estate demand. 

Lutnick's example was Portsmouth, Ohio, where Eisenhower once built America's nuclear fleet, and globalization hollowed out the community. What followed was a $105B project, 9.2 gigawatts of Japanese-funded power, a $100B fab on the site, 20,000-40,000 construction jobs, and 25,000 full-time roles. Multiply Portsmouth by every stranded power municipality sitting on idle transmission capacity, and the compute buildout starts to look like a map of the American interior. This is a localized economic halo.

Texas is another model of reference, according to Bob Frenzel, CEO of Xcel Energy. The state is “building some of the largest transmission infrastructure, and is a destination for data centers. And they’re doing it quite quickly.” Rather than trying to build on China's central-planning model, he said the US should push towards “the Lone Star State model.” 

The Workforce Halo: The Sovereign Employee 

The most urgent takeaways of the week came from a panel discussion with Ursula Burns and Gina Raimondo, moderated by MSNBC's Stephanie Ruhle.

Burns dropped a heater towards the end of that session. “It's very unlikely that a company will unemploy you tomorrow because of AI," she said. "It is likely that you will be displaced over time. If you're sitting still waiting for that job to be safe, I think you're going to be disappointed.” That six-month horizon is the gap between organizational confidence and workforce readiness. Right now, most companies are living in this purgatory.

Raimondo, the 39th US Secretary of Commerce, used a historically powerful analogy to contextualize just how profound AI’s impact is on the workforce: “The so-called China Shock… destroyed manufacturing jobs and whole communities. In that shock, it was a few million people who lost their jobs. 35 years later, we're talking about 20-30 million AI-vulnerable jobs.”

The timeline for this next wave of workforce disruption is far more compressed. 

And she bluntly stated that America's career transition infrastructure is not equipped to handle the aftermath. “What do we have in America for career transition? It's called unemployment insurance, created 100 years ago during the Great Depression. It hasn't changed since. It's not going to work in an AI economy.” She called for new policies, including tax incentives for employers who retrain rather than lay off, reducing payroll-tax friction for entry-level hiring, and outcome-based funding for higher education.

Raimondo then called on businesses to lead the shift, mainly because government entities don’t know the training and education required to keep pace. Then, she noted that governors and mayors can serve as shepherds, supporting new policies and incentive programs because they’re less partisan and more practical. They "actually do work" and, perhaps most importantly, they’re so connected to their communities that they know every CEO they need to connect with. 

The scale of the workforce impact was corroborated by Hugo Sarrazin, CEO of Udemy, who noted that a staggering 92 million knowledge workers will need re-skilling. He is treating it as an existential challenge, and his primary mandate is to transform his organization from an online learning platform to an AI-native platform that upskills users. 

The trust gap is the pressure on all of this. Tom Wilson, CEO of Allstate, and Campbell Brown, Founder of Forum AI, spent time on stage unpacking the disconnect between tech promises and public perception. “There will be a lot fewer people employed doing existing work in not just insurance, but all businesses,” Wilson admitted. “I think the challenge for us is how are we going to use this technology to create more jobs, not get rid of jobs? It’s actually not that hard to use AI to get rid of jobs. What could you do for your customers tomorrow that you don’t do today because it’s either too hard or too expensive? If we don’t do that, trust is going to get even worse.” 

The message conveys intentional urgency for both organizations and individual contributors. Organizations must understand how AI is transforming their industry and how their customers perceive AI use throughout the value chain. Individual contributors must understand how AI is changing the nature of their work and how to use it to gain a productivity advantage. 

The Transportation Halo: Autonomous Cars and Gamified Airlines

In our latest round of predictions, we noted that 2026 would be a critical year for autonomous vehicles, with Waymo among our top companies to watch. In this sector, the AI halo effect ventures beyond bits and into atoms, impacting last-mile logistics, fleet economics, and the human cost of driving. 

Jang of Kindred Ventures framed it sharply: “Self-driving cars are really the first consumer robot that everyone has used in the modern era,” he said. And in his view, the dichotomy between physical and digital AI is false. “Anything in physical AI is instructed and supervised by digital AI. Physical AI is just a natural evolution of what we can do in terms of intelligence, reasoning, planning, and information.”

The commercial front line is already visible. Jang disclosed that Kindred’s portfolio company Nuro has partnered with Uber and Lucid Motors to power the Uber-branded robotaxi fleet. Waymo is expanding. 

The policy layer will determine whether this technology scales successfully across the US. Secretary of Transportation Sean Duffy argued that the federal government should preempt and set the standards, especially if it is the linchpin for global competition. He framed the current NYC blocking of Waymo by Mayor Mamdani as a state’s prerogative, but warned: “The technology that's developed, I would hope, is deployed from America around the world, as opposed to from China around the world. I think we should set the standard.” 

Still, safety should sit at the center of all decision-making, according to Duffy. In highly congested markets like Manhattan, routes can be disrupted by several uncontrollable factors, from random accidents to jaywalkers. He considers himself a ‘Goldilocks’ of sorts, who must balance technological innovation with the safety of passengers. But of course, he brought up the caveat that 40,000 Americans die each year in car accidents, largely driven by drunk driving. Autonomous vehicles can greatly reduce, if not one day eliminate, the impact of this casualty.  

Autonomous driving is one piece of a much larger AI-enabled infrastructure. $12.5B from the infrastructure bill is going into fiber, analog-to-digital conversion, new radar, voice switches, surface awareness, and AI-enabled airspace and flight-management software developed with three contracted companies. 

The most practical, and arguably most unexpected, story from Duffy’s session was the FAA’s gamer recruitment play for air traffic controllers, which speaks directly to the workforce side of physical infrastructure. The DOT polled 250 randomly selected students at the FAA Academy and found that only three were not gamers. That correlation inspired a targeted recruitment campaign aimed at the gaming community. The application portal went live at midnight the night before Duffy’s session. By 7 a.m., it had already received nearly 6,000 applications, with Duffy expecting to cap applications at 8,000 by noon that day. 

“What these gamers are doing on screens—they’re talking, there’s a lot of things going on, they’re used to that,” he said. “That’s actually what you’re doing in a tower in a facility.”

The DOT has compressed the application-to-academy window from roughly two years to 39 days and, by improving candidate fit up front, plans to reduce the washout rate from 30% to 10–15%. On the retention side, Duffy is offering a 20% upfront cash bonus to eligible retirees who stay on, and flagged that internal transfers (e.g., a Chicago controller moving to Miami) pull an experienced controller offline for 12-18 months of airspace-specific retraining, a hidden cost of workforce mobility. His goal is to close the controller shortage by the end of President Trump’s term. 

The DOT is running a public-sector sovereignty play. The federal government is matching adjacent-skill labor to a sovereignty-critical function, controlling American airspace, before the private market can price the labor away or outsource it overseas. The workforce halo of AI is a competition for who gets to own the pipeline, and the public sector is now racing the private sector to build it first. For commerce leaders, the gamer story is a preview of what adjacent-skill recruiting looks like when done with speed and intentionality.

The Investment Halo: Capital Formation and Geo-Economic Influence

Jang of Kindred Ventures predicts that the incoming wave of IPOs, including OpenAI, Anthropic, Databricks, SpaceX, and Stripe, will “relax the liquidity crunch” on early-stage venture and reopen the innovation pipeline. Divesh Makan of Iconiq noted that initial 2026 AI revenue projections of $500B have already been revised to $650B, mainly because the use cases are unimaginable, making adoption harder. “But once you start adopting it,” he said, “you can’t do without it. I’ve never seen technology like this before.”

Gulf sovereign capital is reshaping the geopolitical topology of where AI investment flows. Saudi Arabia, the UAE, and Qatar are building diversified portfolios that include direct partnerships with companies like Meta. Meta Compute is a new division of the big tech company, and over the next few years, the company is investing nearly $600M to support its associated compute and energy needs. “When you think about the size and scale of what is required today in terms of capital, energy, and talent, estimates have it that over $10 trillion of capital will be needed just over the next few years and over 250 gigawatts of power, which is a lot,” explained Dina McCormick, President and Vice Chairman of Meta. “It also requires an enormous amount of engineering talent.” 

Europe is racing to match the depth of private capital. François Villeroy de Galhau, Governor of the Banque de France, called for mobilizing private European savings into equity and venture capital across AI, energy transition, and defense, with France's deficit contracting from 5.8% to 5.1% in 2025 and a 3% target by 2029. Capital allocation, including who funds what and from where, is the question.‍

The Implications for Commerce Leaders‍

The halo effect of AI is already visible across energy, mobility, labor, and capital. Every ring of it is a sovereignty decision in disguise, determining who owns the power, who builds the workforce, who writes the standards for agents, and who ultimately captures the consumer. 

As a result, there are five decisions that are shaping the next wave of innovation:

1. Power: Have you stress-tested your cost structure against the data-center load curve?
   
   
2. Physical AI: Is your last-mile, delivery, and in-store automation roadmap compatible with autonomous vehicles, or are you still optimizing for a human-driver world?
   
   
3. Productivity: Are your employees actually using AI tools with measured ROI, or have you issued licenses without a diffusion strategy?
   
   
4. People: What is your retraining commitment before the first round of AI-driven workforce adjustments?
   
   
5. Trust: Who governs the agents operating on your employees’ and customers' behalf? Can you audit them when something goes wrong?

The leaders who win the next cycle will be the ones who read the halo and position themselves to benefit from the redistribution of sovereignty.

‍