Uber AI Costs Tighten, Vatican Warns, IBM Opens Quantum Foundry

Uber’s AI Token Spending Under Scrutiny

Uber’s chief operating officer, Andrew MacDonald, told investors that the company’s AI token budget is approaching a point where justification becomes a boardroom debate. The comment, reported by Business Insider, follows a year of aggressive model training for routing, demand‑prediction, and dynamic pricing. Those models consume “tokens” – a unit of compute billed by cloud providers – at a scale that now competes with the firm’s traditional driver‑incentive spend.

MacDonald did not disclose the exact dollar amount, but the phrasing signals a shift from growth‑first to cost‑control mode. Uber’s engineering teams have been iterating on large language‑model‑style architectures to improve ETA accuracy and surge‑price elasticity. The token consumption curve, according to internal dashboards, has risen faster than rider‑growth in the past twelve months, prompting senior finance to question the marginal lift versus the token bill.

Vatican’s AI Encyclical Highlights Dehumanization Risks

In a rare technical pronouncement, Pope Leo III released an encyclical that brands opaque AI systems run by a handful of corporations as a catalyst for “new forms of dehumanization.” The document, covered by Variety, warns that when algorithmic decision‑making is hidden behind proprietary code, users lose agency over personal data and societal narratives.

The Pope’s office cites examples where facial‑recognition suites, predictive policing tools, and large‑scale language models concentrate power in firms that can afford the underlying compute clusters. The encyclical does not name specific companies, but the language mirrors ongoing policy debates about algorithmic transparency in the EU and the United States. By framing the issue as a moral concern, the Vatican adds a cultural dimension to the technical discourse that regulators have struggled to articulate.

IBM Unveils First Pure‑Play Quantum Chip Foundry

IBM announced the spin‑off of a dedicated quantum‑chip fabrication facility, marking the first pure‑play foundry focused exclusively on superconducting silicon wafers. The venture is backed by a $2 billion investment from the U.S. Chips Act, a legislative program aimed at bolstering domestic semiconductor capability. According to Futurum Group, the foundry will produce 300 mm wafers populated with superconducting qubits, a step up from the 150 mm formats that dominate today’s research labs.

The new plant promises a supply chain that separates quantum‑chip design from the traditional CMOS fabs that IBM runs for classical processors. By standardizing the lithography steps for superconducting circuits, IBM hopes to cut the time‑to‑fabrication for a 70‑qubit chip from months to weeks. The $2 billion grant also includes provisions for workforce training, targeting a pipeline of engineers versed in cryogenic packaging and low‑noise control electronics.

Industry Context: AI Compute Costs and Quantum Ambitions

Uber’s token‑budget squeeze and IBM’s quantum‑foundry launch illustrate a broader tension in capital‑intensive tech. AI research has entered a regime where model size and training epochs translate directly into cloud‑compute invoices measured in millions of dollars per year. Companies that can amortize that spend across revenue‑generating features survive; those that cannot are forced to reassess.

At the same time, quantum hardware is moving from prototype to volume‑production. The $2 billion Chips Act funding reflects a policy belief that quantum advantage will be a national security asset, much like the early days of microprocessor R&D. IBM’s choice of a 300 mm substrate aligns with the semiconductor industry’s push for larger wafers to improve yield and lower per‑chip cost. If the foundry can deliver repeatable qubit performance, it could accelerate the timeline for error‑corrected quantum processors, a milestone that remains elusive.

What to Watch

Investors will monitor Uber’s next earnings call for any revision to its AI‑spend forecasts and for hints about alternative token‑pricing models. Vatican‑aligned NGOs may track whether the encyclical spurs concrete regulatory proposals on algorithmic transparency in Europe or the U.S. Finally, IBM’s foundry will publish its first production yields by the end of 2027; those numbers will indicate whether the quantum‑chip supply chain can keep pace with the growing demand from academia, defense contractors, and cloud providers.