← All firm reviews
Refinery · Private research

Inflection Ventures: strategic review

23 September 2026 · Deep public-evidence review · Laurie review pending

Inflection Ventures: a distinctive thesis still carried by a small number of people

Review date: 23 September 2026
Research window: 23 March 2025 to 23 September 2026, with historical anchors from 2019 onward
Status: Deep strategic review complete; Laurie review pending

Headline reading

Inflection is trying to own the idea that the best early-stage returns sit in “hard futures”: engineering moonshots at the intersection of computation and the physical world, especially where European sovereignty, unfashionable markets and technical difficulty keep consensus capital away. This is a real position. The firm’s current portfolio, sovereign-computation thesis, first-principles research and low-volume first-cheque model point in the same direction.

The public system is partly effective for that job. It makes Inflection recognisable and gives technically minded founders evidence of how Alexander Lange and Jonatan Luther-Bergquist reason. It offers much less proof of the claimed founder relationship, independent judgement under pressure or commercial consequences of the work. Transmission is concentrated in a modest newsletter audience and Jonatan’s guest appearances. The result is a firm that owns more intellectual territory than public attention.

1. “Hard futures” turns a broad deep-tech mandate into an investment argument

The current site says Inflection backs hard technologies, hard markets and companies that are hard to underwrite, with first cheques of $0.5 million to $2 million and “low velocity and high conviction”. Its September 2026 metadata narrows this further to non-consensus engineering moonshots at idea stage. The portfolio supplies concrete examples across encrypted computation, edge chips, stratospheric platforms, space security, industrial robotics and energy orchestration.

The important move is economic. The June 2026 Hard Futures essay argues that underwriting difficulty thins the bidding crowd and creates a gap between entry price and a possible future. The December 2025 “Heresy” essay locates the same idea inside a critique of concentrated, playbook-driven venture. Repetition teaches the market that Inflection looks for technical and social conditions peers misprice. This is more defensible than generic “deep tech” or “biggest challenges” language.

It is also partly self-sealing. A company being unusual or hard to price does not demonstrate that Inflection has priced it well. The public evidence contains portfolio fit and argument quality, with no realised-return record or independently tested selection advantage. Nucleus Capital explains first-check scientific underwriting and its support model more concretely. NATO Innovation Fund owns the institutional security frame more powerfully. Inflection’s narrower “sovereign computation” synthesis remains the more particular thesis.

This reading would weaken if the portfolio moved towards conventional software, or if investment memos showed “hard” functioning mainly as retrospective packaging. It would strengthen with dated decision artefacts showing which technical or market inflections caused an investment before consensus formed.

2. Hard Futures is a research programme with a point of view, although its architecture still leaks credit

The accessible 18-month universe contains 22 distinct written outputs across the official site and newsletter after syndicated copies are collapsed. The mix is dominated by thesis and technical research, with a smaller amount of founder instruction, firm philosophy and historical digest publishing. Close reading covered investment logic, machine sensing, physical AI, European defence, stratospheric infrastructure, fundraising, fund construction and the 2025 identity change.

The strongest work risks being wrong. “Machine perception” maps underused sensing modalities, estimates capability markets and ends with a venture-company shape. “Free power at high cost” tests wireless power transfer through conversion losses. “Europe’s New Defense” defines 3 venture-backable company types and names structural procurement constraints. “Heresy” makes a contestable case against venture concentration. These pieces expose selection logic rather than decorating investments.

The programme has changed twice. SVRGN’s weekly digest reached at least edition 58 by June 2025, then stopped appearing. The November 2025 site redesign used Janus, classical sculpture, retro-futurist imagery and a custom mark to connect invention with looking backward and forward. In June 2026 the publication itself became Hard Futures. The current site places Rodin beside a robotic arm, gives portfolio companies art-directed thumbnails and uses interactive visualisations inside research. This is unusually coherent creative behaviour for a micro-fund.

The publishing system is still fragmented. Some site research is missing from the Substack archive; 2 newsletter essays are absent from the site; hidden digests remain in the feed; and Kepler, introduced as a public research platform in February 2025, now redirects to the homepage. The current archive makes the work easier to explore, though programme history and ownership are harder to reconstruct. A founder can learn how the team thinks, then take a clear contact route. A repeatable route from a research subject into an event, tool or community is only visible in defence.

3. Partner voices make the thesis believable, while institutional dependence remains high

Alexander defines the house position. He authored the origin story, thesis, venture critique, fundraising instruction and current sensory-AI research. His voice combines personal disclosure with market structure and explicit company shapes. Jonatan specialises it. His writing and appearances connect sovereign computation to cryptography, defence, space, sensing and physical infrastructure.

Two full guest transcripts show that this is more than promotion. In the March 2025 EUVC episode, Jonatan explains the move from crypto to sovereign computation, the scale-resilience-flow framework and why token incentives can distort company building. In the January 2026 Protect Europe episode, he connects 15 European Defense Tech hackathons, frontline input and named investments to diligence. Those appearances give the firm independent hosts, long-form questioning and a practical mechanism unavailable on the homepage.

The same evidence exposes concentration risk. The site lists 5 current people, although only Alexander and Jonatan carry observable public investment arguments. Alex Patow authored engineering and agent-economy work through January 2026 and built Kepler; he is absent from the current team and now identifies with another company on X. Robert S., Rebecca M. and Jon Levin make operations, finance and governance credible, with no public thesis lane located. LinkedIn and X timelines were inaccessible, so broader activity remains unknown.

Jonatan’s defence community adds proof and relationships, yet credit can land on European Defense Tech or a host podcast. The firm benefits when his guest appearances explicitly name Inflection and repeat sovereign computation. It receives less benefit when the activity is encountered as a separate ecosystem project. Full social access or referral data could show stronger integration than public surfaces reveal.

4. Effectiveness is strongest for association and proof of thought, weakest for transmission and verified action

The observable loop is: research or a partner appearance creates a thesis encounter; a founder follows a contact or defence-community route; Inflection may invest or form a relationship; a portfolio example then enters later research. The assets and routes are observed. Content-attributed meetings, investments and outcomes are unknown.

Inflection sits closest to European first-cheque frontier firms including Nucleus, Lunar, Vsquared, OTB and Amadeus, with NATO Innovation Fund as a sovereignty comparison. Nucleus owns a more explicit scientific support model. Amadeus owns longer deep-tech outcome proof and a broader media system. NATO Innovation Fund owns institutional scale and allied-security legitimacy. Inflection more credibly owns the compact intellectual junction of sovereign computation, bits and atoms, and anti-consensus underwriting.

The practical conclusion is preservation, not expansion for its own sake. Hard Futures is doing the right job when it turns a live investment question into a falsifiable technical and economic argument. The position will become materially stronger if those arguments accumulate visible decisions, founder consequences and independent reuse. Private referral data, portfolio interviews, LP feedback or evidence that the publication drives unusually qualified conversations could overturn the current judgement that the ideas travel less far than they deserve.