✨Deep Tech Briefing #80 | First Silicon Photonics Exit since the rise of GenAI & more...
Weekly Intelligence on Deep Tech Private Markets.
Welcome back to Deep Tech Briefing — the weekly space by The Scenarionist where we analyze and discuss the key events of the week shaping Deep Tech Private Markets.
In Today’s Briefing:
Lite Edition
Mini Weekly Recap
Interesting Reading
Full Edition
The Big Idea:
First successful silicon photonics startup exit since the rise of GenAI.
The Key Updates:
The Machine Turns: Why This Week’s “Unsexy” Industrial Tech Headlines Signal the Next Leg of the Buildout.
From 3D ultrasonic sensor, to new US factory to make silicon anodes for energy-dense EV batteries and more.
Signal in the Data:
From Volume to Value: The Industrial Robot Market in 2025
Why the Next Decade Belongs to Intelligent Automation
Deep Tech Power Play:
From space securing institutional funding to Germany’s five-year space spree—plus an EU deforestation delay, a bipartisan push to lift coffee tariffs, and a leaner Air Force small-biz office—this week redraws incentives across defense, trade, and compliance tech.
Dear Friends,
welcome back to Deep Tech Briefing! This week marks edition No. 80! 🎉
It’s been more than a year and a half since the first edition—born as a small weekly project at the request of our most avid readers—and it now counts more than 30K monthly visits!
I am, and we are, proud to follow week after week the crucial yet exciting developments guiding us toward a new wave of reindustrialization.
Grateful to be here—and even more grateful that you’re here. Thank you all! ❤️
Now, let’s recap this week in the deep tech private market world.
Late September 2025 feels less like another AI news cycle and more like an industrial turn.
The constraint now isn’t model architecture; it’s physics—power, interconnect, materials, and manufacturing cadence. Markets appear to be repricing accordingly. The most revealing move of the week came at the physical layer inside the data center, where a U.S. incumbent moved to internalize photonics and adjacent copper links. That signals a shift from chasing raw FLOPS to optimizing I/O, latency, and watts per bit—where the next tranche of efficiency will actually be found.
Inside racks, the roadmaps are getting practical. Optics are migrating closer to compute, short-reach electrical runs are shedding DSP overhead, and serviceability is being treated as a first-class constraint. Performance is increasingly a topology problem: east-west traffic dominates, thermals set the ceiling, and reliability under heat is worth more than a benchmark at room temperature. The metric that matters is cost per delivered gigabit at operating temperature, not the prettiest eye diagram in a lab.
Energy has stepped from backdrop to base layer. New domestic capacity for higher-density electrochemistry is linking up with firm, low-cost power and contracted offtake. The immediate effects are architectural: smaller packs, simpler cooling, tighter packaging. The strategic effect is sovereignty—shorter supply lines, clearer compliance, and better alignment with incentive regimes. In a market where interconnection queues and permitting are the gating items, location and electrons are now part of product strategy.
Automation followed the same script. Factory buyers favored perception stacks that degrade gracefully in dust, glare, and EMI rather than stacks that excel only on clean floors. Safety ratings, EMC hygiene, and mean time to repair are doing more to move purchase orders than exotic range claims. It is a reminder that, in industrial settings, uptime economics beat demo aesthetics every time.
Space continued its progression from manifestos to manifests. Structural work packages were assigned, propulsion plans converged on hybrid profiles to stretch delta-V, and long-duration power options unlocked missions in cold, dark environments. On the ground, idle antenna minutes and fiber backhaul were metered like cloud inventory, compressing time-to-insight for orbital data and turning infrastructure into a marketplace.
Policy signals reinforced the trend. Several states leaned into firm power additions and streamlined industrial permitting; federal demand channels shortened the path for dual-use vendors with credible safety cases. The feedback loop is straightforward: when public buyers pay for delivery instead of paperwork, private capital meets them with factories, tooling, and headcount. Reindustrialization stops being a panel topic and becomes a purchase order.
Enjoy the read!
- Giulia
✨ For more, see Membership | VC Guides | Insights | Rumors | Exit
How Advanced Materials Exhibit Inverse Correlation in Downturns + Toolkit [Downturn Screening Pack] | The Scenarionist
This essay lays out why: from the psychology of flight to quality to the anchoring power of plants, inventories, and multi-year contracts — and when the thesis breaks. It compares three stress periods, shows how VC/PE rotated into hard assets, and closes with operator-level cases and the risk patterns that matter (capex execution, energy sensitivity, export controls).
What you’ll get:
- Why inverse beta appears in materials: collateral, contracts, and qualification moats.
- Historical stress tests across three drawdowns
- Empirical Evidence: Why Assets ‘Fly’ in Fearful Markets.
- Capital flows: how VC/PE reweighted to “real assets” in risk-off.
- Case studies that surface capex, energy, and export-control risks, across batteries, metals, and magnets
- Downloadable Toolkit: Advanced Materials — Downturn Screening Pack
Interesting Reading:
Spending on AI is at epic levels. Will it ever pay off? The Wall Street Journal — the AI-capex building spree (chips, power, DCs) draws dot-com parallels; Bain pegs required output near ~$2T/yr by 2030 to justify it—watch utilization, power deals and real enterprise willingness to pay.
The Pentagon can lead a U.S. manufacturing revival The Washington Post — case for procurement reform: default to commercial processes, curb cost-plus and compliance drag, and use defense demand to crowd in startups and advanced manufacturing capacity.
Big Tech dreams of putting data centers in space Wired — orbital DCs promise 24/7 solar and fewer local externalities, but face brutal realities: radiation, upgradeability, capex-to-launch economics and sovereignty. One to watch for national-security and space-data niches.
A tweezer array with 6,100 highly coherent atomic qubits Nature — a notable neutral-atom scale jump; coherence + control at this count nudges hardware roadmaps from demos toward error-managed, application-relevant regimes.
Meta launches super PAC to fight AI regulation as state policies mount TechCrunch — “American Technology Excellence Project” signals a push to preempt a 50-state patchwork; tens of millions, bipartisan operatives, and a bid to shape model-liability rules via elections.
Nuclear fusion: the race among start-ups to harness limitless, clean energy Le Monde — survey of 50+ players and ~$10B raised, spanning tokamaks to pulsed-power; momentum is real, but timelines and grid-readiness still hinge on capital, regulation and demonstration milestones.
Intel and the rise of U.S. tech nationalization Contrary Research — crisp primer on equity-based industrial policy: the government’s 10% Intel stake as watershed, benefits/risks of state ownership, and how chips, minerals and defense could become strategic shareholdings.
Agrifoodtech investment models need a rethink AgFunderNews — investors argue ag isn’t SaaS: longer horizons, service heavy go-to-market, ROI-proof points; “hot”: robotics/AI; “not”: carbon markets & precision fermentation (for now).
Elon Musk’s xAI offers Grok to the federal government for 42 cents TechCrunch — a GSA deal prices Grok at $0.42 for ~18 months, an aggressive wedge to undercut rivals and seed integrations across agencies.
With AI, chemists create rubber-like materials in record time Phys.org — human-in-the-loop ML speeds design of tough, 3D-printable elastomers; open-sourced tooling hints at broader “AI + automation” discovery loops for specialty materials.