Phew… 2022 was a long year. Inflation is surging, there is a war in Ukraine, capital markets have crashed, and the underlying economy is only beginning to feel the effects of what could be a prolonged downturn. I am not going to provide my macro outlook, as that has been well covered by economists far smarter than I am on the subject. The wisest of the bunch will tell you that we’re in rough waters, there isn’t an easy out, and we don’t know how long this will take to get through (it could be multiple years). For that reason, capital-constrained startups should conserve resources, grow efficiently, and prepare accordingly.
As an early-stage venture investor, I hold a firm belief that you need to keep a level head through any economic environment, boom and bust cycles included. You need to be cognizant of the capital markets and economic headwinds / tailwinds as you invest in and build companies, but it’s been proven time and time again that outstanding businesses are started and built through both up markets and down markets. It’s all about identifying the outliers — unique founders and market opportunities — that have a chance to build an enduring business to survive and thrive through economic cycles over decades. I believe the best VCs invest in those companies early and support them all the way through, regardless of where we are in an economic cycle.
It can be popular opinion in times like these that investors should look for companies that are capitalizing on downmarket themes such as cost savings, counter-cyclical or defensive industries, etc. And while I agree that those themes will likely outperform over the next 1–3 years (or however long this downturn lasts), our strategy is to invest in lasting companies over a decade+ timeline. So it’s most important to find companies that will grow through any market cycle and build an enduring, generational company, as opposed to capitalizing on a temporary trend. These are the companies started by founders that have relentless focus on their long-term mission and aren’t easily distracted by gyrations or volatility in the market. They are the same founders that have avoided getting out over their skis in the past several years of overinflated markets and instead have focused on solidifying product/market fit before investing aggressively in GTM.
For 2023, my investment themes aren’t wildly different than they were last year, but more of a refinement on what I’m seeing as exciting generational opportunities in tech: Data Assembly Line, AI Automation, and Supply Chain (R)evolution.
Each year, I like to take a renewed look at what I consider are first principles for identifying the best early-stage investment opportunities.
- Invest in extraordinary founders with a unique insight and clear vision for how they will define a massive category
- Understand the depth of the problem and viability of the solution
- Develop a thesis for how the company can build defensibility over time
- Define a path to a business model that can become highly profitable
- Take early-stage risk: Welcome GTM, product/technology, and competition risk, NOT founder or market opportunity risk
Theme #1: Data Assembly Line
As I pointed out in my 2022 investment themes, digital transformation means that every company is becoming a software company, and by proxy, every software company is really a data company at its core. Data has become a company’s most important asset. How a company acquires that asset, stores and maintains that asset, protects that asset, and derives value from that asset is mission critical. I call this whole process the “data assembly line”.
The data assembly line has made great progress over the past couple of decades of software development. We now have cloud infrastructure with data storage that can scale nearly infinitely, high performance compute / databases for efficient querying, and even pre-built or open source models to more easily build applications or run sophisticated analytics. As a result, we’re seeing an explosion of real world AI applications that are more accessible than ever before — just look at all of the mind blowing things that OpenAI is doing with ChatGPT and DALL-E 2, bringing generative AI into the mainstream.
However, in 2023 the wide majority of companies are still experiencing real challenges in actually accessing and using the data that they need to solve business problems. The infrastructure is there, and increasingly, the models are there, but getting the right data to the right place at the right time is the hardest part. Due to legacy infrastructure, software, and practices from the preceding decades, most companies are still dealing with data sitting in silos that are difficult to access readily in a secure and compliant way. The tooling needs to improve and the stack needs to mature to help consolidate a fragmented and messy landscape into an efficient data assembly line.
- Data Acquisition & Aggregation: A good foundation is having high quality and differentiated data to begin with. Startups that can creatively acquire and/or aggregate proprietary data assets within industry verticals or market segments will be able to monetize those assets with a heavy premium via API access, analytics products, vertical solutions, etc. Highly profitable Data-as-a-Service (DaaS) and/or SaaS business models exist with data moats when startups acquire high-value data that is difficult to access. The data itself can become the basis of analytics products, AI/ML models, or even full-stack solutions that produce highly valuable companies. Crosslink portfolio company Near Space Labs is an example of a company that is using sophisticated technology to acquire a proprietary and differentiated data asset in the geospatial market that is applicable to several large industry verticals.
- Developer-Friendly Data Infrastructure: Despite the maturation of cloud data storage and compute, there are still meaningful challenges for developers and data engineers to securely access clean data and collaborate cross-functionally. Crosslink portfolio company Vectice is helping solve collaboration and knowledge management for data science teams. Severe challenges also exist in aggregating data from multiple data stores, clouds, or applications across the enterprise, and delivering the data in a clean, digestible format to the user. Developer-friendly experiences with data security and governance capabilities embedded will be winners.
- DataOps: It remains operationally-intensive to execute on the integration, movement, synching, and transformation of data across a company. Low-code or no-code tools for both developers and non-technical business / data analysts to freely integrate, move, synch, and transform datasets will win. A prime example of this is Crosslink portfolio company Syncari, which provides data automation capabilities for RevOps teams synching their GTM data across the entire company.
- Data Applications, Automation, & Analytics: At the top of the stack, I believe there is a strong need for a more integrated data experience for business users to access data insights and automations without unecessary back-and-forth with engineering and data science teams. I’m excited about startups with data analytics and automation solutions that are tailor-made for a given business unit as opposed to today’s horizontal BI or RPA tools — i.e. pre-packaged analytics and automations built for sales, customer success, product, finance, etc. Crosslink has two portfolio companies capitalizing on this for optimizing sales compensation (Forma.ai) and helping finance teams achieve real-time visibility of their financials (Leapfin). I’m also excited about full-stack analytics packages built for vertical industries (financial services, manufacturing, supply chain, healthcare, etc.). Given the abstractions being built for data infrastructure and model development, I can imagine that it will become increasingly easier for business users to build and run their own large-scale data projects, including training sophisticated data / AI models to solve business challenges with less reliance on technical talent.
Theme #2: AI Automation
AI has already moved from “in beta” to “in production” in many markets, though I think we’re still in the early innings of AI automation delivering on its promise. The bulk of the value is still currently accruing to the early adopters, and I believe AI adoption will proliferate widely across the mass market over the next decade. While there is public concern around AI displacing jobs, I firmly believe that in most cases, AI will instead augment human work, allowing people to focus on the higher order bits as opposed to mundane and repetitive tasks that bog them down. This will enable a more productive workforce and company as a whole.
The elephant in the room is that the large tech incumbents, namely Alphabet, Amazon, Apple, Meta, and Microsoft, and highly capitalized venture-backed companies, such as OpenAI, have all been heavily investing into advancing AI research, models, infrastructure and applications for years. They present competitive threats to startups in the arena. I also believe that AI will become embedded into many of the existing software platforms and incumbent product offerings in both enterprise and consumer markets. That said, startups benefit from their ability to focus, and start with solving what may look like small problems to large incumbents. With a blank canvas and nimble approach to product development, startups can address the deepest pain points occurring within enterprises and industries and solve them with pointed solutions powered by AI automation to produce best-in-class products.
It’s impossible to ignore the explosion of generative AI models and applications being showcased by companies such as OpenAI, Google, Stability AI, Cohere, and select others that are bringing 10+ years of research into the mainstream. There is currently a LOT of hype around generative AI, which has been enabled by years of research, open source models, loads of training data, and high performance compute from highly capitalized players. Even in this depressed market, we’re seeing a premium on valuations awarded to companies innovating in this category and there are a flood of startup pitch decks that have the words “generative AI for ‘X’” on them. As with every technology innovation going through this phase of the hype cycle, much of this early activity will be hype vs. value creation. That said, generative AI is a game changer as it leapfrogs your ability as a software developer or company to create sophisticated AI applications from scratch by accessing powerful pre-trained models that are generalized to address common use cases with text, images, etc. This levels the playing field somewhat for the application builders. If you’re a startup or incumbent, you can skip forward years in R&D time and spend to build your application. This is a positive force in accelerating your time to market, but also has negative consequences as it will invite further competition in your space.
While broad-based generalized AI models are super impressive and important the ecosystem, I believe that the most interesting startup opportunities are in developing highly tuned models and intelligent applications that deliver complete solutions to constrained problems.
- AI Automation in Enterprise: AI can be a great tool to solve for manual, laborious, and repetitive tasks bogging down functions in the enterprise. For sales teams, trained ML models can automate the identification of the ideal buyer or champion within a high potential account. Every sales rep can have an AI assistant to automate operational data entry and generate targeted outbound prospecting emails. A customer support rep can use NLP models to sift through text in support tickets to find the highest order customer problems to solve, and even auto-resolve them when possible. Developers can automate code review and debugging with NLP models tuned for a coding language, or even use generative AI to write simple programs altogether as we’ve seen with OpenAI’s Codex or GitHub’s CoPilot. Decision makers and buyers across every function are waking up to the potential for AI to make their teams more efficient and effective, and I’m excited about startups building brand new AI-powered experiences that will reinvent today’s enterprise software stack with intelligent applications.
- AI Automation in Industry: Past innovations in software have typically followed a pattern of starting broad / horizontal, and then verticalizing and becoming more focused over time as the ecosystem matures. As AI adoption moves from early adopters to mass market, I believe we’ll see more great AI companies built within vertical industries. I’d argue that AI is an even better fit to be applied vertically than traditional vertical SaaS solutions. With AI, you are combining data with sophisticated models, and training those models to improve at accomplishing a targeted outcome. By training AI models with high quality data to solve an industry problem, you can produce a highly effective and targeted solution for customers that is unmatched by traditional vertical software. There is also a high degree of defensibility and “winner take all” dynamics associated with vertical AI companies — more customers generate more data, which train better AI models, which produce better products, which attracts more customers... We have seen success with this model in several Crosslink portfolio companies — Overjet in the dental industry, Inscribe in financial services, and Arturo in insurance. I’m excited for what I believe will be a boom time for startups building AI solutions to solve core problems within industry verticals.
Theme #3: Supply Chain (R)evolution
One of the painful learnings from the COVID-19 pandemic is that our global supply chain is extremely brittle, and has in result been deeply fractured over the past several years. Supply and demand shocks occurred simultaneously during the pandemic, geopolitical tensions have been on the rise across key centers of our global economy, and we continue to see the impacts of climate change. In result, we have experienced supply constraints in key inputs across our global economy, including energy, agriculture, and semiconductors, and the vulnerabilities of our global supply chain have exposed themselves.
It makes sense how we have gotten here. Price competition amongst manufacturers eventually led to mass specialization and globalization of trade to drive supply and manufacturing costs down. Over time, this created a fragmented, complex, and interconnected global supply chain that relies on a complicated logistics network to get supplies sourced, products manufactured, and end products delivered to their destination. While this has worked well enough during more stable times, the recent pandemic combined with geopolitical unrest exposed cracks to the foundations. Just look at what happened to automobile manufacturing — there are ~20k parts in the average car that are sourced from all over the world, so you can understand why disruptions to the supply chain caused automakers to cut production by an estimated 10M units in 2021. In response, manufacturers across almost every industry are scrambling to diversify their supply chains, consider onshoring manufacturing where feasible, and invest in new technologies and processes to rebalance their supply chains.
I believe that startup opportunities in supply chain right now are endless. I’m most interested in 1) Companies that close the loop among the various constituents of the existing supply chain and logistics network, and 2) Companies that can remove inefficient links from the current complex supply chain and logistics network.
- Data Visibility: Given the complexity and fragmented nature of the current supply chain and logistics networks, there is a need for better data visibility for all constituents in a transaction. Much of the existing supply chain and logistics players are still operating with technology from the 1980s that leave them in the dark when it comes to both internal and external data. I’m excited about startups that tap into 1st party and 3rd party datasets to introduce better data visibility into the discovery, pricing, and process status between parties. Crosslink portfolio company Telegraph is a great example of this concept in the freight rail market.
- Simplifying Transactions & Multiparty Collaboration: Supply chains are inherently a chain of multiple parties conducting B2B transactions. As digital transformation has brought more workflows online, I believe there are startup opportunities to streamline the collaboration between these parties, who often still communicate via phone, email, or even fax machine (!!). Similarly, fintech infrastructure has come a long way and there are opportunities for startups to simplify transactions, automate laborious financial operations, and embed financing products to improve cash flow dynamics for both sides of a transaction. Crosslink has invested in financial enablement for the freight brokerage industry (Denim), an end-to-end industrial supply chain solution (GoExpedi), and a technology-enabled B2B marketplace for the lumber industry (Yesler), each with elements of this theme incorporated in their businesses.
- Advanced Manufacturing: 3D printing / additive manufacturing, AI + IoT, robotics, and other technologies applied towards advanced manufacturing are an exciting way to introduce further automation and control over the manufacturing process. The cost of deploying robotics and sensors in a manufacturing setting have reduced dramatically and combined with the advanced capabilities of AI make it possible to dramatically improve the efficiency of a manufacturing line. Similarly, 3D printing / additive manufacturing is a game changing technology that has matured to deliver on production-grade quality. Crosslink portfolio company OPT Industries is an example of how combining advanced software with additive manufacturing technology can greatly reduce design cycles and rapidly scale production capacity. These technologies when deployed effectively allow companies to consolidate their manufacturing processes, bring more of the process in-house, and reduce reliance on third party suppliers offshore. This can have multiple benefits: 1) More control over the manufacturing process and end product quality, 2) Improved manufacturing efficiency (i.e. cost savings), 3) Less reliance on third party suppliers offshore, and 4) Reduced logistics spend, coordination, and risk of shipping delays.
In a challenging macro environment, there are real benefits to being a startup. You’re nowhere near saturated in your target market, so there are a lot of potential customers. Product is still nascent enough that it can be iterated, augmented, or pivoted altogether to meet the market where it is. You can readjust your financial plan, keep your team small, and focus on more capital efficient growth. You don’t need to worry about controlling the narrative with public market investors who may sell your stock if you make meaningful adjustments to product, growth plans, or team construction. In many ways, this environment is refreshing as a startup and early-stage investor as it heightens our sense of focus. We can focus without distraction on the top level mission of building exceptional products that solve problems for customers and build a lasting company through economic cycles.
I’m excited to continue the mission with our portfolio companies, work collaboratively with my favorite co-investors, and meet the next generation of category defining companies. Cheers to 2023!