Alpha School's billion-dollar experiment has never been graded

LAST WEEK, First Lady Melania Trump walked a red carpet at the White House flanked by a five-foot-tall humanoid robot. The machine, built by Figure AI, addressed an audience of international first spouses in eleven languages before tottering back down the hall and disappearing. The spectacle was part of Trump's Fostering the Future Together summit, at which the First Lady asked the room to imagine a humanoid educator named "Plato" that would provide personalized instruction to every child in America. It was a vision straight from the pitch deck of one particular school, a school whose 11-year-old student had sat in the First Lady's box at the State of the Union just weeks earlier, and whose co-founder has donated more than $2 million to Republican candidates and school-choice PACs since 2023, according to the Washington Post.

That school is Alpha School, and it is no longer a fringe experiment. With more than 1,000 students across 22 campuses and plans to reach 35 locations by fall (including new openings in Chicago, Atlanta, and Puerto Rico), Alpha has become the most visible, most politically connected, and most polarizing educational venture in America. Its model is audacious in its simplicity: students spend two hours each morning learning core academics entirely through AI-driven software on laptops, with no textbooks, no homework, and no teachers. The classrooms are instead staffed by "guides," adults instructed not to teach but to motivate. By lunchtime the laptops close and students shift into afternoon workshops covering everything from public speaking to pottery. Alpha calls this model "2 Hour Learning" and reckons it can deliver academic outcomes that surpass those of traditional schools by a factor of two.

The expansion has been aggressive even by Silicon Valley standards. In July 2025, Alpha acquired key assets from Higher Ground Education, the former parent company of Guidepost Montessori, once the world's largest Montessori network, fast-tracking the opening of ten new campuses. Tuition ranges from $40,000 per year at the Austin flagship to $75,000 in San Francisco; the Chicago campus opening this fall in the former GEMS World Academy building will charge $55,000. Education Secretary Linda McMahon visited Alpha's Austin campus in September 2025, praising its use of AI as a model for the nation, and President Trump signed an executive order in April 2025 directing McMahon to prioritize federal funding for AI in education. Alpha student Everest Nevraumont, an 11-year-old TEDx speaker and three-time Texas state history champion, joined Melania Trump in the First Lady's box at the 2026 State of the Union.

Alpha School may be pointing toward something genuinely important about what traditional education gets wrong. Before it is allowed to define that future, however, it has to prove, independently and transparently and at scale, that its claims are true. So far, it has not.

The core insight behind Alpha's model is not controversial among cognitive scientists: a single teacher lecturing 30 students at the same pace, regardless of whether some are two years ahead and others a year behind, is an extraordinarily inefficient way to transmit knowledge. Adaptive software that meets each student at precisely their level of mastery, identifies gaps before moving forward, and provides immediate feedback is, in principle, a better way to handle the mechanical work of learning facts and procedures. Anyone who has watched a bright child sit through a lesson they mastered weeks ago, or a struggling child fall further behind because the class moved on, can see the appeal.

Some families report transformative results. Parents describe children advancing multiple grade levels within months, developing genuine enthusiasm for academics, and (perhaps most strikingly) asking not to take summer breaks. The mastery model allows students to race ahead in subjects where they have aptitude and spend more time in areas where they struggle, without the social stigma of being held back or the boredom of being held in place. At Alpha's San Francisco campus, lead guide Carson Lehmann describes adults freed from the grind of delivering curriculum and able to focus on what they find most rewarding: mentoring students and providing emotional support.

The afternoon workshops are, by most accounts, the strongest part of the program and the thing Alpha is actually selling when it talks about "giving kids their time back." Students build sailboats and run mock businesses. They learn financial literacy and coding, take pottery classes, and go on outdoor expeditions. At least one workshop per day requires physical activity. If Alpha's academic claims hold up, the argument that children are better served by two hours of hyper-personalized instruction and five hours of real-world skill-building than by six hours of one-size-fits-all lectures is not absurd. It may, in fact, be correct.

But it also invites an obvious question Alpha has never satisfactorily answered: if the AI tutor genuinely produces two-times-faster learning in two hours, why stop at two hours? A child could, in theory, do a second session in the afternoon, learning a second language, music theory, advanced writing, or the kind of deep interdisciplinary thinking that elite schools have always prized. Instead Alpha treats the academic day as something to be compressed and gotten over with so the real day can begin. For a school that charges up to $75,000 a year, the choice to do less rather than more with a supposedly superior tool is a curious one. It raises the possibility that the product being sold is not accelerated education but accelerated childhood: a lifestyle proposition for affluent families, wrapped in the language of academic rigor.

Alpha's headline claims are familiar to anyone who has encountered its marketing: students learn two times faster, classes rank in the top 1% nationally, and growth rates are 2.6 times faster than peers on NWEA's Measures of Academic Progress exam. These are striking numbers, and they rely entirely on internal analyses. The underlying data has not been independently reviewed by outside researchers, nor has Alpha published results in any peer-reviewed journal. CNN reporters requested access to Alpha campuses and interviews with co-founder MacKenzie Price for months; both requests were declined.

The testing methodology raises its own questions. Justin Reich, director of MIT's Teaching Systems Lab, has pointed out that the MAP assessment Alpha uses is primarily administered in public schools across the country, drawing students from a wide variety of economic backgrounds. Comparing the test scores of children whose families pay $40,000 to $75,000 in annual tuition against a general population that includes Title I schools is not an apples-to-apples comparison. A more informative benchmark would measure Alpha's students against their peers at schools like Harker, Dalton, or Menlo, where families spend comparable sums and where decades of college placement data exist. Alpha has not offered such a comparison. Northwestern's Liz Gerber has described the model more bluntly: self-directed learning with Montessori principles, repackaged with a technology veneer.

The independent research that does exist on AI-driven education is mixed. A peer-reviewed study published in January 2026 in the Proceedings of the National Academy of Sciences found that the use of Khan Academy, one of several third-party apps Alpha licenses alongside IXL, Membean, Mentava, and MobyMax, was linked to modest math gains, but only as a supplement to teacher-led instruction, not as the primary mode of teaching. A 2026 Stanford review of more than 800 academic papers found mixed results and no clear evidence that fully AI-driven instruction consistently outperforms traditional methods.

The charter school record is also telling. Alpha-affiliated organizations have applied for public charter status in six states, the only realistic pathway through which the model could reach families who cannot afford private tuition. Five of those states rejected the applications. The Pennsylvania Department of Education described the instructional model as "untested" and lacking alignment with state academic standards; the proposed staffing plan listed one Head of School, five senior guides, ten junior guides, and a single special education guide for an initial enrollment of 500 students. Only Arizona approved.

The governance structure merits separate scrutiny. Public charter applications, corporate filings, and independent reporting have documented a network of for-profit entities hired for core services: 2 Hour Learning supplies the adaptive learning platform, Trilogy Enterprises manages financial services, Crossover Markets recruits educators, and a separate LLC provides general and administrative services. These entities share founders, leadership, and operational DNA with Alpha itself. The co-founder billed as Alpha's public face is MacKenzie Price; her husband Andrew is the CFO of Trilogy, Crossover, Ignite Technologies, and ESW Capital, the investment firm of Alpha's principal and primary funder, Joe Liemandt. The founders have also established other schools using the same instructional model under different names (GT School, NextGen, Novatio, Unbound, and Valenta), all licensing the 2 Hour Learning platform.

Liemandt is not an educator by background. He is a software operator whose career illuminates the logic behind Alpha's design. In the early 1990s he dropped out of Stanford to found Trilogy Software, becoming the youngest self-made member of the Forbes 400 before the age of 30. Through ESW Capital, an investment arm of Trilogy, he spent the next two decades acquiring hundreds of aging software companies, replacing U.S. employees with overseas contractors at scale, and monitoring their productivity through a system called WorkSmart, described in a Forbes investigation as a tool that took screenshots of workers' screens every few minutes and tracked mouse clicks and keyboard strokes. Since 2022, Liemandt has invested $1 billion of ESW's cash flow into education, according to a profile in Colossus. His net worth is estimated at $6.2 billion. The product he calls the best he has ever built, Timeback, is designed to scale Alpha's model to a billion children on sub-$1,000 tablets.

The operational pattern across Liemandt's career is consistent: identify a high-cost knowledge worker, replace the most expensive elements of that role with software and lower-cost labor, monitor the system digitally, and extract the efficiency gain. That pattern does not make Alpha wrong. But it does suggest that the model's design is shaped at least as much by an operator's instinct for cost compression as by a pedagogy.

Alpha School may be building something genuinely valuable. The mastery-based model has real merit, the afternoon workshops are creative and well-liked, and some children are thriving in ways their parents never expected. The fundamental premise — that the traditional six-hour school day is a relic of an industrial-era labor schedule, not a pedagogical optimization — is one most parents intuit every time they watch their child grind through busywork at 9 p.m.

The questions worth asking are not whether AI belongs in education (it almost certainly does) but whether the evidence matches the expansion, whether the governance structure serves families or founders, and whether a model that costs more than the median American mortgage payment can credibly claim to be a path toward universal access. These are separate questions. They deserve separate answers, and Alpha has hitherto provided none of them.

If a child can genuinely master a year of math in five months, the implications extend far beyond one private school network. It would mean that millions of American children are spending thousands of hours in classrooms not because they need the instruction but because the system was never designed to let them leave. It would mean that the most valuable resource in a child's life, unstructured time to build things, explore interests, and develop the kind of agency no curriculum can teach, is being consumed by a model that optimizes for compliance rather than competence. Every parent who has watched a gifted child grow bored, or a struggling child grow ashamed, recognizes the waste. That is a future worth building carefully, verifying honestly, and, if the evidence holds, embracing without apology.

But a billion dollars and a White House invitation are not evidence. The children are the evidence. And so far, no independent researcher has been given access to verify the results. ■

For more, join 75,000 subscribers getting tech's favorite brief.