OpenAI and Anthropic dominate as US AI startup funding hits US$150B
Despite the record haul, most of 2025’s AI funding went to a few big names. Marketers and tech teams should take note
US-based AI startups pulled in a record US$150 billion in 2025, surpassing the previous high of US$92 billion in 2021. This year’s frenzy of deal-making was headlined by megadeals such as OpenAI’s US$41 billion raise and Anthropic’s US$13 billion round. Investors included SoftBank, Andreessen Horowitz, Thrive Capital, and Tiger Global.
For B2B marketers, infrastructure buyers, and startup founders in the AI space, the headline number signals momentum. But as capital concentrates in a few players, the sustainability of this growth and its implications for the broader ecosystem deserve closer scrutiny.
This article breaks down the record-setting 2025 AI funding wave, the infrastructure risks behind the numbers, and how marketers and platform teams can prepare for what’s next.
Short on time?
Here’s a table of contents for quick access:
- Who’s getting the money? Spoiler: it’s not early-stage startups
- Infrastructure spending raises new red flags
- FinOps tools are gaining traction in GenAI cost management
- What marketers should watch next
Patrecia Meliana
Who's getting the money? Spoiler: it's not early-stage startups
At first glance, US$150 billion sounds like a golden age for AI startups. But more than one-third of that capital went to just two companies. OpenAI raised US$41 billion, and Anthropic brought in US$13 billion. The majority of investor interest focused on foundation model developers building massive, general-purpose AI systems.
This funding wave favors established players over newer ones. We still don’t know how much of the capital reached seed or Series A startups. The lack of transparency in stage allocation makes it hard to gauge whether new entrants are being supported or squeezed out.
If investor sentiment turns, these large bets could become vulnerabilities. The market’s optimism is currently high, but it may not last if macroeconomic concerns or infrastructure inefficiencies emerge.
Infrastructure spending raises new red flags
Some VCs are already urging caution. Running large-scale GenAI models is expensive, and inference costs—the cost of producing outputs from models—can eat up 80 to 90 percent of AI operating budgets.
Despite raising billions, many AI startups are facing GPU underutilization rates between 70 and 85 percent. That means compute resources are sitting idle while costs continue to mount. It’s a warning sign that some startups may struggle to sustain operations, even with large cash reserves.
Investors are advising founders to build buffers now, before the funding window narrows. If companies don't shift toward efficiency, future rounds may come with tougher terms or dry up entirely.
FinOps tools are gaining traction in GenAI cost management
For marketing and engineering leaders deploying or evaluating GenAI, cost control is no longer a back-office concern. It’s central to long-term competitiveness.
FinOps platforms are now essential in helping AI teams reduce cloud spend and improve efficiency. These platforms offer multiple cost-saving tactics:
- Dynamic scaling adjusts compute use based on real-time demand, reducing GPU costs by 40 to 70 percent
- GPU pooling allows AI workloads to share GPUs, increasing utilization across teams
- Token optimization reduces the number of text units processed by models, cutting inference costs by 20 to 40 percent
The FinOps sector is now valued at US$5.5 billion with a 34.8 percent compound annual growth rate. As AI applications expand, these tools are becoming critical for maintaining margins and investor confidence.
What marketers should watch next
The surge in funding is impressive, but the real story lies beneath the surface. Here’s what to track:
- Funding concentration signals platform risk
Most capital is flowing into large foundation model companies, not niche or vertical solutions. Marketers should diversify vendor bets and avoid over-relying on a few players.
- Demand better cost transparency from partners
If your AI vendor isn’t optimizing GPU use or token loads, you may be paying a premium for inefficiency. Platform teams should push for metrics before signing new contracts.
- Explore FinOps capabilities early
Whether you’re building internal tools or working with external AI providers, financial operations are now a key part of infrastructure planning.
- Prepare for market shifts
The funding landscape could change quickly. If investor enthusiasm cools, your partners may pivot or restructure. Stay flexible in your go-to-market plans and vendor relationships.
