Video summary

AS TECNOLOGIAS QUE VÃO MUDAR O MUNDO NA PRÓXIMA DÉCADA | Market Makers #368

Main summary

Key takeaways

Technology

Market Makers #368 — Gui Peremúter (Deeptech Investing)

The video is an episode of Market Makers #368 featuring Gui Peremúter, a deeptech venture investor. The discussion focuses on which technological shifts are likely to matter most over the next decade—especially AI, semiconductors, robotics, space-related infrastructure, and biotech/life sciences—and how to invest in them.

1) AI as “general-purpose technology” + acceleration

  • AI is framed as moving beyond novelty and becoming a general-purpose / infrastructure technology—akin to how electricity, writing, or the transistor reshaped the world.
  • The next decade is presented as a period of rapid acceleration, driven by the interplay of semiconductors + AI.
  • AI is expected to act as a platform that boosts many industries, including:
    • biotech
    • robotics
    • manufacturing
    • agriculture
    • education
    • services
  • AI is expected to become embedded in everyday devices, where “Do you use AI?” becomes as normal as “Is there Wi‑Fi?”—including:
    • phones
    • TVs
    • clocks
    • thermostats
    • glasses

2) “Investing in inevitability” (risk-neutral thesis building)

Gui’s investment approach emphasizes building a thesis by neutralizing controllable uncertainty and focusing on what is believed to be inevitable, rather than purely speculative.

Examples of “inevitabilities” used to guide themes:

  • Rising life expectancy → greater need for diagnostics and therapies for aging-related diseases, including:
    • neurological conditions
    • oncology / cancer
    • ischemias
    • autoimmune diseases
  • Ongoing urbanization, including automation reducing rural labor → impacts agriculture efficiency, infrastructure, mobility, energy, and security

Core idea: instead of forecasting the next year, focus on 10-year curves—summarized by the market adage:

Overestimate the next 12 months; underestimate the next 10 years.

3) Why predictions fail—and how to think about 10 years

The episode argues that short- and mid-term forecasts often fail because even experts routinely misjudge technology timelines.

Historical examples mentioned include:

  • hype-cycle patterns where expectations were wrong (e.g., multimedia / robotics)
  • computing resource mispredictions such as:
    • “not more than five computers”
    • “640KB is enough”

The proposed investor mindset for the next decade includes expecting underappreciated but real change, such as:

  • intelligent automation/robots (not necessarily humanoid; optimized for specific tasks)
  • space as an investable economy, with near-term targets like Moon + Earth orbit, not only Mars

4) Space economy as a new investment vertical

The next decade is described as opening space-linked investment verticals, including:

  • off-Earth resource prospects (e.g., “mining off-Earth”)
  • industries in lunar orbit / low Earth orbit / geostationary orbit
  • equipment and services:
    • spacecraft parts
    • satellites
    • probes
    • rockets

Gui also suggests that access to space exposure may expand through public listings—referencing the general expectation around a SpaceX IPO.

5) Deeptech investing method: staged validation + specialized expertise

Deeptech VC is contrasted with liquid markets:

  • Liquid markets reprice instantly.
  • VC relies on milestones / checkpoints.

Checkpoint concept

Investments are structured around staged proof:

  • fund early proof/prototype progress (e.g., “execute the first 10% in ~6 months”)
  • if milestones fail, the startup may effectively run out of trajectory
  • if execution improves, raises and valuation typically increase

Specialization in the investor ecosystem

Because deeptech domains are complex, the investment ecosystem is described as specialized by field (e.g., biotech, robotics, astronautics, AI), limiting debate mainly within niches.

6) Brazil deeptech paradox (good science, weak commercialization)

Gui argues Brazil performs relatively well in academic output, citing:

  • high-quality articles and citations
  • a position around 14th–15th among 194 countries

However, Brazil is said to rank poorly in business/investment competitiveness due to:

  • bureaucratic constraints
  • fiscal issues
  • labor constraints
  • operational difficulties

Key contrast

  • In deeptech, the underlying knowledge/product is often globally competitive from birth (e.g., new molecules/proteins/circuits/algorithms).
  • But local ecosystem frictions can make it harder to turn breakthroughs into scalable companies.

Gui also highlights Brazil-focused initiatives, including:

  • the organization of a Deeptech Summit at USP, bringing together investors, entrepreneurs, and government representatives, with expansion toward Latin America participation.

7) Portfolio examples (tech domains with concrete products)

The episode includes company stories to illustrate deeptech themes and productization.

AI / AR (human-machine interface)

  • Control Labs (in portfolio), acquired by Meta Used as evidence that deeptech themes can reach consumer products (AR glasses interface advancements referenced).

Defense / automation

  • Anduril Mentioned with founders/tech lineage (linked to Palmer Luckey and defense modernization narrative). Used to illustrate power-law outcomes: one “winner” can represent a large share of VC returns. Also notes geopolitical/governance constraints and the need for processes like CFIUS for strategic investments (U.S. context).

Agritech + robotics in the field

  • Halter (New Zealand → investments expanded globally)
    • uses animal collars and vibrations to control cattle movement remotely
    • aims to optimize water/pasture and reduce labor (replacing herding-dog labor)
    • described as subscription-based with strong retention (claims: no customer loss reported)

Biotech / medical diagnostics (AI + computer vision)

  • Cytoval (as spelled/mentioned in subtitles: CT O V A L E / Citoval)
    • focuses on sepsis / septicemia diagnosis speed
    • problem: current blood tests take 12–36 hours, while mortality risk rises quickly
    • solution: a device uses computer vision + AI to classify blood samples by risk within ~10 minutes, analyzing cell response to air compression

8) AI governance and risk framing (agents + permissions)

Gui dismisses the “AI gains consciousness and plots against humans” narrative as overly romanticized.

Instead, he highlights a practical risk:

  • when AI systems become autonomous agents with greater permissions, they can cause harm

He uses an access-control analogy, such as:

  • giving someone the master password for destructive capabilities

He also ties risk to software reliability—especially coding mistakes—including comparisons to blockchain/DAO failures where bugs can scale into major events.

9) Market bubbles + how they impact deeptech VC

Gui acknowledges VC “bubbles,” comparing current AI/tech euphoria risk to historical analogies such as:

  • the dot-com era
  • electrification-era infrastructure buildouts by large firms

Impact on deeptech VC:

  • early-stage VC is less exposed to liquid-market repricing than later-stage exits
  • deeper in the startup stack, valuation sensitivity is described as lower
  • risk is greater for later-stage companies aiming for IPO / M&A timing

10) Company-building philosophy: deeptech output is physical/material too

Across book recommendations and the “material world” theme:

  • even if the core is software, deeptech ultimately depends on the physical world
  • this includes materials and supply chains such as copper, steel, and sand/silicon inputs needed to make technology real

Main speakers / sources

  • Gui Peremúter — guest; deeptech investor; founder of Grids Capital; author/columnist; presented the “investing in inevitability” thesis framework
  • Thiago Salomão — host; Market Makers founder/CEO; introduces and questions Gui

Original video