The latest from 8VC
Dirt: A Love Story
The concept of building holds mythic status in the world of new venture creation, and the mythos often obscures the difficulty. In construction, there are no such illusions. Construction is a $2 trillion U.S./$13 trillion global industry where automation has barely scratched the surface of possibility. Beneath that surface is dirt—unfathomable amounts. For new housing subdivisions, commercial real estate developments, data centers to power the AI boom, highway expansions, and countless other projects, American abundance begins with dirt.
These projects can require moving dirt on a scale of hundreds of thousands of cubic meters. For perspective, 100,000 cubic meters is 40 Olympic swimming pools. Today, this requires enormous CapEx (excavators ranging from $80,000 minis to $500,000 for the largest) and skilled operators ($100,000/year fully burdened). By 2030, the average skilled equipment operator will be over 46 years old (vs 42.9 today), and not enough young people are replacing them. Meanwhile, construction as a whole is nearly 500,000 workers short of meeting 2025 demands. [1]
Introducing Bedrock
At 8VC, we look at this scale of challenge through the nexus of what is newly possible through technology, and the extreme edges of talent. Advances in AI and reinforcement learning have created incalculable opportunity in construction and other physical industries, but doing it well is very hard. There are few teams with plausible talent, and only one with both the talent and experience.
The Bedrock team achieved some of Waymo’s defining breakthroughs, including freeway driving, all-weather operation, systems reliability, and sensor/compute architectures for autonomy—all while meeting rigorous safety requirements. CEO Boris Sofman was Waymo’s Head of Perception and led Waymo Via (autonomous trucking and delivery solutions), helmed the Waymo One freeway launch, and was previously CEO and founder of Anki. CTO Kevin Peterson worked alongside Boris leading perception, software reliability, and release for Trucking & Freeway, and was previously co-founder and CEO of Marble (acquired by Caterpillar, where he headed Caterpillar Robotics).
While road autonomy is hard and incredibly data-intensive, construction is a different beast. Boris and Kevin’s experience solving the toughest and most persistent problems in autonomy, combined with their roles in establishing a winning culture, have given them a unique radar for where you can move quickly given the right constraints.
Bedrock’s value proposition is making general contractors (GCs) more productive by winning and completing more projects, faster, and more safely. Bedrock’s deep technical investments enable a low-lift approach, retrofitting standard equipment and multiplying human inputs. They are deployed onsite moving dirt with top-tier partners like Sundt Construction (a $3.5 billion/year GC founded in 1890), as well as Zachry Construction, Champion Site Prep, and newly signed Capitol Aggregates. Through these partnerships, Bedrock is gaining the insight and repetitions needed to build unmatched workflow software for autonomy. This translates not only to moving dirt efficiently, but also to completing all the work that precedes building new structures. The ultimate goal is reliable, general-purpose construction autonomy at scale.
Technical Approach
The Bedrock tech stack builds on a number of trends and developments that will allow construction robotics/autonomy to turn the corner from money pit to money printer:
- Hardening of foundational autonomy technologies (system architecture and design, controls, evaluation, perception systems, etc)
- Ubiquitous, affordable sensor hardware
- Advances in models and ML over the past five years (specifically end-to-end learning, a breakthrough approach for problems that have historically been too nuanced/complex for engineered solutions).
Bedrocks’ offering comprises three layers:
1) A low-cost autonomy kit that integrates with all OEM functions, from basic manipulation to advanced electronics.
2) End-to-end learning. Waymo’s environments provided extensive evidence for this approach. The ability to learn on entire trajectories and generalize is imperative for navigating an unpredictable environment like a bare construction site, with fluid, efficient, lifelike motions.
3) Comprehensive workflow software for single-task definition (i.e. with a CAD file as the input) and multi-task planning and orchestration (ingesting schematics, reports, geotechnical data, etc).
%20(1).png)
The Road Ahead
Excavation is an ideal start because site prep is foundational, and excavators are among the most utilized and hardest to operate machines. Solving excavation first opens the door to easier functions (loading, dozing) and coordination of multiple machines. Beyond automating specific tasks, we can now ask “what does ‘Devin for construction’ look like?” Could an agentic system go from high-level plans, to coordinating fleets and teams, to performing long-horizon tasks?
First-generation AI services plays were a key theme for 8VC Fund V. Fund VI takes these learnings into the physical realm, where there are likewise numerous businesses that can suddenly be made many times more productive. Imagine a quarry that produces 24/7, or a subcontractor that works day and night at much lower cost. There are trillions of dollars of value to capture by building as fast as possible. For companies like Sundt, the goal can move from excellence to being the best in the world: saying yes to every meaningful project and reducing inefficiencies to levels previously unimaginable in heavy industry.
These technologies will help realize the promise of more and better infrastructure and lower cost of goods, affecting millions and ultimately billions of lives. In this sense, Bedrock is an investment in a renaissance of American development. Breaking ground is the step towards solving the housing crisis, industrial onshoring, smart agriculture, livable cities, and sustaining data infrastructure to power all of the above.
Red tape and anti-change politics will be tough to budge (and it would be naive to think otherwise), but we are betting that politics will ultimately come down to what is best for hundreds of millions of Americans. Furthermore, these are secondary concerns without proving that the work itself can be done. Given the dire shortage of operators, wide adoption of construction AVs will still require all hands on deck. Far from replacing jobs, this domain targets work that isn’t being done today, and will create massive economic activity. Bedrock is on the vanguard of what we might term “American Maximalism”: fulfilling the promise embodied in our land and citizenry.
.jpg)
Bedrock x 8VC
Today, Bedrock announced their $60 million Series A, led by 8VC. This represents our most ambitious bet in autonomy to date, echoing Boris and Kevin’s vision. There are clear tailwinds: advances in models and hardware, a fantastic wedge case, market pull stemming from the labor crisis, and rapid retrofit opportunities enabling faster scaling. That leaves the team as the deciding factor, and every signal we gathered reinforced that Bedrock’s is in a category of one.
With unrivaled founders, a critical bottleneck to American prosperity, and technical advancements itching to be proven on a vast scale, we knew it was time to back up the proverbial excavator. We are privileged to support Bedrock in solving a foundational class of problems, and ensuring that it is always time to build.
[1] https://www.machinerypartner.com/blog/labor-shortages-in-construction-solutions-for-recruitment-and-retention
The American Development Renaissance: Our Investment in Bedrock Robotics
The concept of building holds mythic status in the world of new venture creation, and the mythos often obscures the difficulty. In construction, there are no such illusions. Construction is a $2 trillion U.S./$13 trillion global industry where automation has barely scratched the surface of possibility. Beneath that surface is dirt—unfathomable amounts. For new housing subdivisions, commercial real estate developments, data centers to power the AI boom, highway expansions, and countless other projects, American abundance begins with dirt.

The American Development Renaissance: Our Investment in Bedrock Robotics


Drew Oetting On Building An $8 Billion Venture Capital Firm To Invest In Startups That Are Fixing Our Broken World
Drew Oetting is one of the biggest forces providing the financial fuel this new generation of fast-growing, super-sized startups need to make it. His venture capital firm, 8VC, has invested in startups like Unlearn, Chaos, Tome, and Ushur.
Announcing Our Investment in Outset
Ask anyone who’s ever begun a term paper the night before it’s due: there are stark tradeoffs between research time and research quality. This is especially true of primary user research, one of the fundamental ways large enterprises can ensure their products resonate with customers. Historically, this has required a choice between user interviews, which are high-fidelity but slow, manual, and expensive, and surveys, which are scalable, cheaper, and faster, but often result in unrefined, low-signal data. Although there have been some strong companies on the survey side, e.g. Qualtrics and Medallia, this uneasy compromise has always persisted—until now.
Charles Srisuwananukorn (Together AI) Fireside Chat
We were thrilled to feature Charles Srisuwananukorn from Together AI at January’s Chat8VC. Charles is the Founding Vice President of Engineering at Together AI, where he leads the company’s work on AI infrastructure and clusters. Previously, he was Head of Applied Machine Learning at Snorkel AI and held engineering roles at Apple. He studied Computer Science at Stanford and has helped steer Together from an early contributor to open-source AI to a full-stack infra platform.
Joe Chen and Jonathan Shen (Upwork) Fireside Chat
We were thrilled to feature Joe and Jonathan from Upwork at March's Chat8VC in San Francisco. We covered their journey from teams like Google Brain and Cruise, and their own startup, to leading AI efforts at Upwork—building Uma, a suite of specialized LLMs powering workflows for freelancers and clients across the platform.
Clear Eyes, Fuzzy Joins, Can’t Lose: Announcing Our Investment in Structify
Human-quality workflows need human-quality data, an axiom that has only grown truer in the AI-first enterprise. However, access to complete, high-signal data remains a limiting factor, given steep data provider fees, inflexible schemas, AI hallucinations, and scattered, inconsistent, and mutating sources. Customers don’t need to be data scientists to recognize shovel-ready datasets, but if they need to be data scientists to generate them reliably, data will always be rate-limiting.