Private aviation runs on tight margins and tighter schedules. The AI tools Jet.AI built to optimize both placed the company in an unusual vantage point: watching production inference workloads run against real operational constraints, before the data center power shortage became a mainstream story. Mike Winston, investor and founder of Jet.AI (NASDAQ: JTAI), built those tools inside an operating aviation business and drew from them a conclusion that now anchors two public companies: the constraint binding the AI infrastructure buildout is power, and the gap between available supply and projected demand will persist for years. That conclusion informs the February 2025 agreement to transfer Jet.AI’s aviation operations to flyExclusive, the data center development pipeline being assembled through the Convergence Compute joint venture, and the $138 million SPAC raised through AI Infrastructure Acquisition Corp. (NYSE: AIIA). For investors trying to understand Jet.AI’s trajectory, the aviation chapter is where the thesis actually originates.
From Jet Token to Jet.AI: A Sequence With a Logic
What happened at Jet.AI between 2016 and 2025 reads, from the outside, as a series of technology pivots. The company began as Jet Token, a blockchain-based private aviation startup founded by Mike Winston, CFA, whose prior career had run from equity research at Credit Suisse First Boston through five years as a portfolio manager in merger arbitrage and event-driven investing at Millennium Partners. Regulatory constraints closed off the blockchain model’s commercial path. The company rebuilt around AI tools for aviation: agentic booking software, route optimization for fuel and carbon efficiency, dynamic pricing for charter operations. Each change tracked external conditions. Each stage also produced information the next depended on.
What Building Aviation AI Software Actually Reveals
The tools Jet.AI developed for private aviation required real compute at operational scale. Agentic booking software coordinates availability, pricing, and scheduling across multiple aircraft against a customer base with variable and often short-notice demand. Route optimization requires running real-time models against weather, airspace, and fuel data. Dynamic pricing models consume compute at a rate that scales with transaction volume and prediction complexity.
Running those workloads inside an operating aviation company (not in a research environment, in production, against real cost constraints) produces a specific kind of knowledge. The compute requirements of operational AI are higher than they appear from the outside. The power requirements of compute at scale are higher still.
“Through building AI tools for aviation, we saw firsthand the scale of transformation AI would bring,” Winston said in an April 2026 interview. “That led us to data centers, where the infrastructure opportunity is significant. Given my background in real estate finance and telecom, it was a natural transition. Today, we’re extending that into power generation using aero-derivative engines, another area with strong underlying demand.”
That insight came from operating a business where AI was a production tool, measured against real cost constraints.
The Power Problem, Quantified
The constraint Winston identified by operating inside aviation AI is now visible across the broader market.
The U.S. Department of Energy estimated data center electricity consumption at 176 terawatt-hours in 2023. Analysis by Alderman & Co. projects that figure could reach 580 TWh by 2028. That would put data centers at between 6.7% and 12% of all U.S. electricity. Grid interconnection queues in some U.S. jurisdictions now run eight to 10 years, measured from the point of application.
New gas turbines from major manufacturers are not closing that gap fast enough. Contact GE Vernova today for an LM6000 order and the delivery window runs three to five years minimum. GE Vernova CEO Scott Strazik said in early 2025 that the company expected to be largely sold out through the end of 2028 by that summer. Siemens Energy reported that more than 60% of its U.S. gas turbine orders that year were linked to AI data center demand. Mitsubishi’s newer turbine blocks ordered in 2025 may not ship until the 2030s.
The practical solution for data center operators who need power now is the aero-derivative gas turbine: units built around retired commercial jet engine cores, modified for stationary generation. ProEnergy has sold 21 of its PE6000 units to just two data center projects: more than one gigawatt of combined bridging power. Each unit produces 48 megawatts and can be operational within 30 days of delivery. ProEnergy was quoting 2027 availability when GE Vernova’s order book had already closed into 2028 and beyond.
The Aviation Industry as an Early Observer
The CF6-80C2 turbofan engine, the core unit that ProEnergy overhauls for its ground-based power systems, was widely used on Boeing 767s and Airbus A310s. Approximately 1,000 of these engines are expected to retire from commercial aviation service over the next decade. The supply is quantifiable, the retirement schedule is predictable, and the companies with operational knowledge of aviation hardware were positioned to recognize the secondary market forming around those cores.
Jet.AI was an aviation company with AI tools and capital markets literacy. That combination produced an earlier read on the intersection of retiring aviation hardware and data center power demand than financial analysis alone typically generates.
The competition for aero-derivative turbines has since created cross-sector friction that Alderman & Co. analysts Ryan Kirby and Joseph Lakaj documented in March 2026: aero-derivative units share a near-identical manufacturing base with commercial flight engines, relying on the same specialized castings, high-temperature alloys, and precision forgings. A large data center order for turbines now directly competes with engine deliveries for new commercial aircraft. Boeing and Airbus are both navigating extended delivery timelines driven in part by engine shortfalls. Two industries are pulling on the same supply chain, and the aviation sector is both a contributor to that constraint and, through companies like Jet.AI, a beneficiary of it.
The flyExclusive Transaction and What It Unlocked
The agreement to transfer Jet.AI’s aviation operations to flyExclusive removed the operational complexity that had kept two structurally different businesses inside a single public vehicle.
flyExclusive takes the Citation and HondaJet fleet and the private aviation customer base. The combined platform has the scale to extract returns Jet.AI’s aviation division could not reach independently. Jet.AI shareholders receive flyExclusive (NYSE American: FLYX) equity alongside their retained JTAI position. The post-close version of Jet.AI carries no fleet, no pilots, and no charter operating costs.
What remains in JTAI: the Convergence Compute joint venture with Consensus Core Technologies, targeting one gigawatt of data center capacity across three campuses in North America; a $5 million economic interest in a special purpose vehicle anchored by SpaceX and xAI equity; and the 49.5% economic stake in the AIIA sponsor.
On June 1, 2026, Glass Lewis issued a “FOR” recommendation on the flyExclusive merger. Glass Lewis is one of two proxy advisory firms whose research institutional investors consult as a standard checkpoint before shareholder votes. The special shareholder meeting is scheduled for June 11, 2026. Approval requires an affirmative vote from a majority of all outstanding shares. Institutional participation is essential to clearing that threshold.
Public markets tend to undervalue companies that operate across two structurally distinct businesses. Aviation and AI infrastructure attract different investors on different time horizons. Separating them into distinct listed vehicles removes the valuation friction that a mixed balance sheet creates.
AI Infrastructure Acquisition Corp.
AIIA raised $138 million in its October 2025 IPO. Its mandate is to identify and close a business combination in data center infrastructure or AI, a focus the company describes as “ship to grid.” As of early 2026, management confirmed active engagement with several targets.
The connection to JTAI runs through sponsor economics. SPAC sponsors typically receive 20% of post-IPO equity as founder shares plus warrants exercisable at $11.50. Jet.AI’s 49.5% position in the AIIA sponsor entity means that if AIIA closes a qualifying business combination, nearly half the sponsor economics flow back to JTAI shareholders. The stake was carried at $17.23 million on Jet.AI’s balance sheet as of Q1 2026, and the company reported $13.5 million in cash with no debt.
Winston has positioned the infrastructure bet across two independent paths: an organic buildout through Convergence Compute and an acquisition vehicle through AIIA. The structure means not every outcome depends on a single execution.
Winston’s Background and the Pattern It Reveals
Winston joined Credit Suisse First Boston in 1999 on a telecom research team that Institutional Investor Magazine ranked first, at the start of one of the largest infrastructure capital cycles of the modern era. Five years at Millennium Partners followed, co-managing a $1 billion merger arbitrage and event-driven book through Catapult Capital Management. That discipline produces a specific habit: determine what an asset is worth if the market-moving event does not occur, then price accordingly.
The data center power thesis runs through that same lens. The demand is documented: grid interconnection timelines, turbine manufacturing lead times, and hyperscaler capex commitments are all public record. The question event-driven analysis poses is not whether the demand is real but whether the specific positioning captures the value before it prices in. Winston has spent a career in disciplines that reward being right about that second question.
He founded Sutton View Capital in 2012 after departing Millennium Partners. The firm advised one of the largest academic endowments in the world and co-led successful activist litigation against the Dole Foods board, securing a 35% increase in total consideration for shareholders. The CFA credential, the Institutional Investor ranking, the Columbia MBA: the credentials are institutional. The career decisions have been independent. Jet.AI and AIIA are both built outside established platforms, on conviction about where specific structural conditions point.
Where the Risk Lives
AIIA has a standard SPAC window of 18 to 24 months from its October 2025 IPO. No business combination has been announced. The clock is running, and trust account mechanics create real deadline pressure regardless of whether the acquisition market cooperates on the same schedule.
Convergence Compute has three of four development milestones complete, with power studies and permitting underway across its three campus sites. Construction, equipment procurement, and customer acquisition follow. The financial returns depend on those campuses being built, leased, and stabilized. Each step carries execution risk appropriate to a company of JTAI’s current scale.
The supply constraints that make the thesis credible are also the supply constraints that make execution difficult. Developer competition for turbine delivery slots, permitting capacity, and project financing is intensifying as more capital chases the same infrastructure gap.
The observational logic that runs from aviation AI tools to data center infrastructure holds up as an account of how Winston read the market. Whether Jet.AI can execute against it before the supply advantage narrows is what the next 18 months will determine.
Disclosure: This article discusses Jet.AI, Inc. (NASDAQ: JTAI) and AI Infrastructure Acquisition Corp. (NYSE: AIIA). Readers should conduct their own due diligence before making investment decisions. This piece reflects publicly available information and does not constitute investment advice.
