Alex Kolicich, 8VC | AI & Defense Tech Renaissance

Market Dynamics & Valuation Uncertainty

• Investors face "peak uncertainty" regarding where value capture will accrue within the AI stack, ranging from hardware and foundation models to application layers and customer systems.
• Satya Nadella's thesis that all winners in a technological phase shift are decided within the first two to three years is driving aggressive moves by incumbents like Microsoft and Google to avoid being left behind.
• Public software companies are currently valued at approximately 6.5x forward revenue, returning to a 10-year average, while early-stage AI startups are seeing multiples of 100x to 200x revenue.
• Private market valuations are "priced to perfection," with AI startups raising hundreds of millions at valuations far exceeding public comps, a trend sustained by an oversupply of capital in mega-funds.
• Venture capital funds are incentivized to write large checks (e.g., $100M+) rather than smaller ones, as larger check sizes allow them to deploy more capital while limiting dilution to roughly 20%.
• The "AI capabilities gap" refers to the current limitation where LLMs are "almost good enough" for production but still suffer from hallucinations and errors, particularly in complex tasks like text-to-SQL conversion.
• Alex Koloseich predicts a near-term future of "man-machine symbiosis" rather than full autonomy, where AI handles 80% of work under human supervision to bridge the reliability gap.
• The market is currently at or near the "peak of inflated expectations" on the Gartner hype cycle, suggesting a future "trough of disillusionment" as revenue fails to catch up to infrastructure investment.
• Sequoia Capital data indicates that aggregate capital expenditure (CapEx) on AI hardware vastly exceeds the revenue generated by AI companies, signaling a potential mismatch in value creation.
• OpenAI's recent funding round valued the company at $157 billion; however, Koloseich notes that even at a 15–20x forward earnings multiple, the valuation depends heavily on margin structures not yet fully transparent.

Investment Strategy & The "Tart" Theory

• Koloseich advises AI and AI-adjacent startups to raise capital aggressively every three to four months, citing the Eugene Kleiner quote: "The time to take a tart is when it is being passed."
• He warns that the current capital spigot will eventually turn off, normalizing multiples and making fundraising significantly harder if companies wait.
• Venture managers risk becoming "one-trick ponies" reliant on pre-AI business models, with many expected to retire or be fired over the next decade for failing to adapt investment theses.
• CRV's return of fee-bearing assets from its growth fund is viewed as an anomaly reflecting a shrinking pool of fundable companies rather than a systemic trend in venture capital.
• The total addressable market for venture-scale software opportunities has shrunk by an estimated 60–70% since 2021.
• Mega-funds are currently absorbing risk in the foundation model layer, with no significant write-offs yet reported from companies like Inflection, encouraging continued aggressive deployment.
• Koloseich suggests that "non-AI-first" startups will be priced to converge toward 7x forward revenue multiples if they grow efficiently, creating a widening valuation gap with AI-native companies.
• A new market opportunity is emerging for sovereign wealth funds to execute fund-wide or position-wide secondary sales for 10-year-old venture funds lacking liquidity exits.

Defense Sector Transformation

• The defense industry is undergoing a shift due to the war in Ukraine, which has moved warfare toward small-group infantry, drone autonomy, and high-precision strikes, exposing gaps in traditional defense primes' capabilities regarding software and AI.
• 8VC has incubated multiple defense companies (Anduril, Saronic, Epirus, Chaos Industries) by acting as co-founders who provide shared resources for government engagement and large upfront capital.
• Defense startups now compete successfully against traditional primes (e.g., Raytheon) by leveraging technical excellence in areas where primes lack agility, such as rapid software deployment and autonomy.
• The industry sentiment has shifted from hostility to celebration, with "defend" startups becoming a "crown jewel" of Silicon Valley, a reversal from the era when Google withdrew from military contracts.
• SpaceX and Palantir are cited as vanguard companies that validated the defense sector, creating a "well-worn path" for new entrants and changing the government's procurement culture.
• 8VC emphasizes that successful defense founders typically have prior military service or experience at companies like SpaceX, Palantir, or Anduril, rather than being "outsiders."
• The defense market is predicted to create "tens" of new primes rather than hundreds, given the budget constraints relative to the total economy, suggesting a more concentrated exit landscape.

Structural & Contractual Changes in Defense

• Firm Fixed-Price (FFP) Contracts: SpaceX pioneered the shift from "cost-plus" (where risk is on the government) to FFP (where risk is on the contractor), aligning incentives for efficiency and cost control.
• Other Transaction Authority (OTA): Replaces bureaucratic FAR contracting (900-page contracts) with streamlined, one-page agreements that allow for faster decision-making and shorter procurement cycles.
• Defense Innovation Unit (DIU): Provides early-stage funding (e.g., $10M–$30M) for prototypes and engages with private investors to de-risk solutions before they enter the formal federal budget cycle.
• Bridging the "Valley of Death": New programs exist to match private sector funding with government grants to support manufacturing capacity before full procurement orders are approved, reducing the time to market.
• Contracting Risks: Koloseich notes that startups should anticipate antitrust scrutiny; companies like Microsoft and Amazon are acquiring teams via "acquihires" or licensing models to secure foundational layers without triggering full M&A approval processes.

Internal Movements & Leadership Anomalies

• Koloseich highlights a highly irregular trend of top leaders departing AI companies (specifically OpenAI) pre-IPO, noting that while mass exodus is common after an IPO, it is unprecedented during a company's "heyday."
• Specific leadership departures include Ilya Sutskever leaving OpenAI to launch a new AI lab with $1 billion in funding, and the mass exodus of the "Devin" team (Cognition AI), which Koloseich cites as a surprise.
• These departures have led Koloseich to reconsider his bullishness on application-layer startups; he now believes there is significant room for new "supervisor software" companies (like Devin) to disrupt incumbents by managing autonomous agents.
• The "Devin" example specifically shifted Koloseich's view from incumbents capturing all value to the possibility of new, agile workflows replacing traditional software systems.
• Koloseich expresses caution regarding OpenAI's future leadership continuity due to the instability of its top talent, questioning if the company will remain the dominant leader.
• The current environment is described as "unstable" for VCs and founders, with rapid changes in model performance, scaling laws, and team composition occurring quarterly.
• Scaling laws for LLMs are showing decreasing returns, requiring order-of-magnitude increases in compute for sublinear performance gains, raising questions about the near-term path to AGI.