Journal
AI Capital Is Getting More Selective. What That Means for Founders and Investors
Sep 28, 2026 · Donoghteh

Recent fund moves from Lightspeed and Bessemer point to a more focused AI investment market, where execution, local insight, and measurable customer outcomes matter.
# AI Capital Is Getting More Selective. What That Means for Founders and Investors
Artificial intelligence remains one of venture capital’s biggest magnets. But the latest fund announcements suggest the conversation is evolving: the question is no longer simply whether investors will fund AI. It is where capital can produce enduring value, and how quickly investors expect to put that capital to work.
Two recent moves illustrate the shift. Lightspeed is targeting a $250 million India fund focused on early-stage AI, while Bessemer Venture Partners announced $5.75 billion across two new funds to invest across the AI stack. The headlines are about large pools of money. The underlying signal is about focus.
Capital is abundant, but strategy is becoming more specific
Lightspeed’s new India vehicle is half the size of its predecessor and is designed around a shorter investment period. According to TechCrunch, the fund had already secured commitments for 80% of its $250 million target, with investments expected to begin soon. The firm’s thesis centers on AI opportunities in India and Southeast Asia, especially the application layer, where a deep software talent base and established technology-services ecosystem can help companies turn models into useful products.
Bessemer’s $5.75 billion raise points to breadth at a different scale. The firm said it will invest across the AI stack, with $1.75 billion allocated to seed and early-stage investing and the remainder to growth opportunities. Rather than a single bet on one model or interface, this approach spans the infrastructure, tools, and applications that enable AI adoption.
These strategies are not contradictory. One narrows its geographic and thematic lens; the other reserves capacity across layers and stages. Both reflect a more deliberate effort to match capital to a defined opportunity set.
Why fund size is not the same as deployment advantage
A large fund can support ambitious companies and follow-on rounds, but it is not an advantage on its own. The important questions are how quickly the firm can identify high-quality opportunities, what expertise it brings after investing, and whether it can distinguish durable businesses from products benefiting mainly from a temporary wave of attention.
A smaller, focused fund can move quickly and build deep local knowledge. A larger platform can support companies through multiple stages and invest across complementary parts of an ecosystem. In either case, disciplined deployment matters: capital should follow evidence of customer need, product performance, and a credible path to sustainable economics.
For founders, this means the identity of a fund may matter as much as its headline size. Does the investor understand your market? Can it help recruit, sell, navigate regulation, or finance the next stage? Does its time horizon match the company’s development cycle? These are practical questions, not cosmetic ones.
The opportunity in AI applications is increasingly local
The India opportunity highlighted by Lightspeed is especially instructive. Building globally competitive frontier models requires exceptional compute, research talent, and capital. Yet the value of AI is not limited to training the largest model. Local-language interfaces, industry-specific workflows, customer support, financial services, healthcare operations, and software development can all benefit from applying models to concrete problems.
Markets differ in language, regulation, distribution, purchasing behavior, and available data. Companies that understand those conditions can create products that are difficult to replicate simply by releasing a general-purpose model. The opportunity is not to put AI on top of every existing product. It is to redesign important workflows so that AI makes them measurably faster, cheaper, or more accessible.
For founders, the bar is therefore shifting from “we use AI” to “we reliably improve this outcome.” The strongest products will make the improvement visible to customers and repeatable in production.
What founders should demonstrate now
**A real customer problem.** Explain the costly or frustrating workflow your product improves, and who owns that problem.
**Evidence of outcomes.** Show time saved, error reduction, higher conversion, improved service levels, or another metric that matters to the buyer. Demonstrations are useful, but sustained use is stronger evidence.
**A path to dependable delivery.** Investors and customers want to know how the product handles quality, latency, privacy, security, and failure. Human review may be appropriate, but it should be an intentional part of the system rather than an unexplained patch.
**An economic model that can endure.** Model costs, support requirements, and pricing should make sense as usage grows. If gross margins depend on a future reduction in inference costs, make that assumption explicit.
**A defensible route to market.** Distribution, trusted relationships, proprietary workflow data, integration depth, and domain expertise can be more durable than access to a widely available model.
What investors should measure beyond the AI label
Investment teams can look past model choice and evaluate whether a company has repeated usage, clear willingness to pay, and a credible expansion path. They can test whether the product is embedded in an important workflow, whether customers can switch easily, and whether performance improves with real deployment data.
They should also distinguish between a compelling prototype and a business that can operate reliably. Evaluation should include the full cost of serving customers, human intervention, integration work, and the consequences of inaccurate outputs. These details can determine whether adoption compounds or stalls after a pilot.
A more mature AI market rewards proof
The recent fund announcements do not suggest that investor interest in AI is fading. They suggest that capital is being organized around sharper theses, specific stages, and a broader understanding of where value may accrue. The next phase will reward firms that can deploy thoughtfully and founders who can translate technical capability into customer outcomes.
For startups, the practical message is clear: build for a consequential problem, measure the improvement, and earn trust in the workflow where your product operates. For investors, the opportunity is to back teams with the focus and execution discipline to turn AI’s promise into lasting businesses.
The headline may still be AI. The differentiator is increasingly what the technology helps a customer accomplish.
Sources
- TechCrunch, “Lightspeed targets $250M for new India fund, focusing on early-stage AI,” September 24, 2026: https://techcrunch.com/2026/09/24/lightspeed-targets-250m-for-new-india-fund-focusing-on-early-stage-ai/
- TechCrunch, “VC firm Bessemer now has another $5.75B to invest in (what else?) AI,” September 23, 2026: https://techcrunch.com/2026/09/23/vc-firm-bessemer-now-has-another-5-75b-to-invest-in-what-else-ai/