Bloomberg Technology

Pari Singh on giving hardware its software moment

Pari Singh· Founder & CEO of Flow Engineering at Flow Engineering
·~10 min·English·Bloomberg Technology
AgentsRoboticsAI CompanyBusiness Strategy
TL;DR

Flow Engineering's Pari Singh argues AI is collapsing hardware design cycles the way it already collapsed software's, turning multi-year waterfall development into a continuous, agent-driven loop, a shift investor Roelof Botha backs as durable because hardware is central to most of the world's most valuable technology companies.

01Core Thesis

Software's Curve Comes to Hardware

<strong>AI already cut software's design cycle from two weeks to two hours</strong>, and Singh's bet is that hardware is the next domain to compress.

Flow is doing for hardware engineering what AI has already done for software engineering.

— Pari Singh, Bloomberg Technology
Key Insight
The analogy hides its own precondition. Software's leap worked because code is text an AI can read, write, and test in a loop. Flow's real bet is that CAD, simulation, and requirements can be made just as machine-legible, because the curve only follows once hardware's artifacts become an AI-readable loop.
02Why Flow Exists

The 90/10 Problem

<strong>Singh says 90% of an engineer's time goes to "digital manual labor" and only 10% to real innovation</strong>, the grind Flow aims to delete.

10% of the time would go into the actual innovation and the architecture, and 90% of the time would go into digital manual labor.

— Pari Singh, Bloomberg Technology
Key Insight
Notice what the number reframes. Flow is not selling faster CAD, it is selling the deletion of the work between the CAD, the re-entry, the lookups, the reconciliation. That is why Singh frames the unit of value as iterations per year rather than features in a tool.
03How It Works

From Waterfall to Continuous Verification

<strong>Flow's agents turn a 2-3 year waterfall into continuous verification</strong>, listening to each CAD, simulation, or Git change and working out its impact.

companies go from this traditional waterfall model where a design cycle might be two or three years to this continuous verification cycle where engineers are able to make changes in CAD, simulation, GIT

— Pari Singh, Bloomberg Technology
Key Insight
The load-bearing word is impact. Any tool can log a change; the hard part in hardware is propagating it, because one change can cascade through thermal, structural, and regulatory constraints at once. Owning that dependency graph is the real product.
04The Team Model

When AIs Live on the Org Chart

<strong>Singh counts AI as headcount</strong>, a 50-person team of 25 humans and 25 agents that scales to 100 or 200 with 50 humans.

a 50 headcount team with 25 humans to a 100 or a 200 headcount team with 50 humans, and AI is taking up real roles

— Pari Singh, Bloomberg Technology
Key Insight
This is the quiet redefinition in the interview. If AI is headcount, then output per human, not output per employee, becomes the metric a hardware startup optimizes, and a 50-person company can carry the surface area of a 200-person one without the coordination tax.
05The Investor's Lens

Hardware Was Never a Side Bet

<strong>Of the 100 most valuable companies, 35 are tech and 26 of those depend on hardware</strong>, according to Botha.

If you actually look at the 100 most valuable companies in the world, 35 of those are technology companies, and 26 of those 35 are either predominantly hardware businesses or have meaningful components of their business that depend on hardware development

— Roelof Botha, Bloomberg Technology
Key Insight
Botha is pre-empting the obvious objection, that hardware is hard, low-margin, and out of fashion. By anchoring to the most valuable companies, he reframes Flow not as a bet on a comeback, but as tooling for the segment that already won.
06Traction

Even the Incumbents Are Retooling

<strong>Even incumbents are retooling</strong>, with a General Motors division and the Volkswagen-Rivian venture rebuilding core development around Flow.

They were pre AI, and they're completely retooling their core development practices with flow and AI at the center.

— Pari Singh, Bloomberg Technology
Key Insight
The tell is system of record. A faster tool gets swapped out; the system of record for requirements and verification becomes the place the design functionally lives, which is the position that compounds and is hardest for a rival to dislodge.
07The Durable Bet

Reshoring Is a Tailwind, Not the Thesis

<strong>Botha calls "industrializing America" a real tailwind but warns it is not enough</strong>, because administrations change and tides turn.

I don't think it's enough, though. You know, government administrations can change and, you know, those tides can turn.

— Roelof Botha, Bloomberg Technology
Key Insight
This is the discipline of a long-horizon investor: never underwrite a thesis on a policy you cannot control. Botha is signaling the bet survives a change of government, which is exactly what a founder raising at a $750M valuation needs the market to believe.