The AI industry is growing very fast, but it is also spending money at record levels. Top AI companies are burning more cash than ever before, and early-stage startups are using up $100 million much faster than they did a decade ago.
Even though revenues are rising and about $270 billion was invested in AI in 2025, the gap between earnings and spending is getting bigger. In this article, we look at AI Company Burn Rate Statistics 2025-2026, including how much companies are spending, how quickly startups are burning cash, and what this means for growth, risk, and profitability.
Key Stats: AI Company Burn Rate (2025-2026)
- AI startups raised ~$270 billion in 2025, over 52% of global VC funding.
- Total AI infrastructure spending reached ~$400 billion in 2025, more than double 2024 levels.
- OpenAI burned ~$8.5 billion in 2025 and could spend ~$115 billion by 2029.
- xAI is burning ~$1 billion per month with only ~$500 million in annual revenue.
- AI startups now burn $100 million in ~3 years, about 2x faster than a decade ago.
- The average AI startup spends $5 for every $1 of new revenue, far higher than SaaS (~$1.6).
- Around 90% of AI startups fail, with 85% shutting down within 3 years.
- 95% of enterprise GenAI projects fail to deliver measurable ROI.
- Startup runway has dropped to ~12 months, increasing funding pressure.
- GenAI companies are expected to burn $100+ billion combined by 2027.
Top AI Labs: AI Company Burn Rate Snapshot
Top AI companies are growing quickly, but they are also spending huge amounts of money. In this section, we look at how much companies like OpenAI, Anthropic, and xAI are spending to build and run AI.
Even though their revenue is increasing, their costs, especially for computing and research, are even higher. This makes AI a very expensive industry where companies need to invest a lot before they can make profits.
OpenAI
OpenAI stands out as one of the most prominent examples of high cash burn in the AI industry. In 2025, the company surpassed $20 billion in annualized revenue, yet continued to report significant losses. In the first half of 2025 alone, OpenAI recorded $4.3 billion in revenue while spending $6.7 billion on research and development, resulting in a cash burn of $2.5 billion. For the full year, total cash burn is estimated at approximately $8.5 billion.
The long-term outlook is even more substantial. According to Reuters, OpenAI projects a cumulative cash burn of around $115 billion through 2029. The year-by-year projections are as follows:
| Year | Projected Cash Burn |
| 2026 | ~$17 billion |
| 2027 | ~$35 billion |
| 2028 | ~$45 billion |
| 2024-2029 total | ~$115 billion |
A major reason for these high costs is computing. OpenAI is expected to spend about $10 billion to 11 billion on compute alone in 2026. This is because running its services for around 810 million weekly users requires a huge amount of energy, about 47.2 gigawatt-hours every day.
Right now, OpenAI spends about $1.69 for every $1 it earns, and its losses are expected to stay high, at around 57% of its revenue through 2027.
To put this into perspective, researchers at Deutsche Bank estimate that between 2024 and 2029, OpenAI will spend more money than Uber, Tesla, Amazon, and Spotify combined spent before they became profitable.
Anthropic
Anthropic is also spending heavily, but it appears to be managing its growth more carefully. In 2024, the company burned about $5.6 billion. For 2025, it is expected to burn around $5.2 billion while generating about $9 billion in revenue.
One major challenge has been higher-than-expected costs for running its AI models on servers from Google and Amazon, which were about 23% above estimates and put pressure on profits.
At the same time, Anthropic’s revenue has grown very quickly from $1 billion in late 2024 to about $14 billion by early 2026.
Its profitability has also improved, with gross margins rising from -94% in 2024 to 40% in 2025, though still slightly below its own targets. Importantly, the company expects its losses to shrink significantly, with its burn rate projected to drop to just 9% of revenue by 2027, showing a much more efficient path compared to OpenAI.
xAI (Elon Musk)
xAI has the highest spending compared to the money it makes. By mid-2025, the company was burning more than $1 billion every month, which adds up to about $13 billion for the full year. However, its expected revenue for 2025 is only around $500 million. This makes it the least efficient among the major AI companies in terms of spending versus earnings.
To support this level of spending, xAI has been trying to raise about $9.3 billion through a mix of debt and investment, with more than half of that money planned to be spent within just three months.
OpenAI and Anthropic are also spending heavily, but their revenues are much higher. OpenAI is expected to burn about $8.5 billion in 2025 on roughly $20 billion in revenue, while Anthropic may burn around $5.2 billion on $9 billion in revenue.
In comparison, xAI’s spending is much higher relative to its revenue, with losses exceeding 96%, making it the most capital-intensive among the top AI labs.
Burn Rate Comparison: Major AI Labs
| Company | Monthly Burn Rate | Annual Burn (2025 Est.) | Annual Revenue (2025) | Loss Ratio |
| OpenAI | ~$708 million | ~$8.5 billion | ~$20 billion | ~57% |
| Anthropic | ~$433 million | ~$5.2 billion | ~$9 billion | ~58% |
| xAI | ~$1 billion | ~$13 billion | ~$500 million | ~96%+ |
Industry-Wide AI Company Burn Rate Statistics
The rapid growth of AI is being fueled by an equally massive surge in spending across the industry. In this section, we look at how venture capital, infrastructure investment, and startup burn rates are scaling together.
From record-breaking funding rounds to rising costs for compute and data centers, the data shows that AI companies are not just raising more money; they are also spending it faster than ever.
VC Capital Deployment vs Burn
The AI sector is consuming capital at a historically unprecedented scale.
- In 2025, AI startups raised about $270 billion, which was over half (52.7%) of all global venture capital, for the first time, AI took the largest share.
- Total spending on AI infrastructure (like data centers and hardware) reached around $400 billion in 2025, up from about $168 billion in 2024.
- Generative AI startups alone raised $49.2 billion in just the first half of 2025, already more than the total for all of 2024.
- Major AI companies like OpenAI and Anthropic raised about $80 billion in 2025, more than double what they raised in 2024.
- 58% of all AI startup funding came from very large deals of $500 million or more.
- Even though only about 15% of US venture capital funds focus on AI, they account for around 40% of the total money raised.
Overall, this shows that AI is attracting huge amounts of investment, but also requires massive spending to grow.
ALSO READ: AI Startup Funding Statistics 2025-2026
Speed of Capital Consumption
AI startups are using up money much faster than before. Data from Silicon Valley Bank shows that startups founded around 2022 spent $100 million in about 3 years, around twice as fast as startups did 10 years ago.
At the same time, they are also growing revenue faster, reaching $100 million in revenue in just about 2 years. This means both spending and growth are happening more quickly.
On average, a Series A AI startup spends about $5 to earn every $1 in new revenue. This is much higher than traditional SaaS companies, which usually spend about $1.60 for every $1 earned. Because of these high costs, about 50% of US enterprise software startups may need to raise more money or sell their business within the next 12 months.
Global AI Infrastructure Spending By Category in 2025
AI infrastructure spending in 2025 is extremely high, showing how expensive it is to build and run AI systems. Total spending is around $400 billion, with the largest share going to data center construction at about $236 billion.
A big portion is also spent on compute hardware like GPUs and servers ($100 billion to 150 billion), along with power and cooling ($80 billion to 100 billion) and networking and storage ($50 billion to 70 billion).
| Infrastructure Category | Estimated 2025 Spend |
| Total AI infrastructure capex | ~$400 billion |
| Global AI data center build | ~$236 billion |
| Compute hardware (GPUs/servers) | ~$100–150 billion (40–50%) |
| Power & cooling | ~$80–100 billion |
| Networking & storage | ~$50–70 billion |
McKinsey & Company estimates that companies may need to invest as much as $5.2 trillion in data centers by 2030 to keep up with growing AI demand.
ALSO READ: AI Infrastructure Spending Statistics
AI Company Burn Rate Benchmarks by Stage
Burn rates vary widely by startup stage, and understanding these benchmarks is key to evaluating growth and efficiency. In this section, we break down how spending typically scales from early-stage startups to more mature companies, along with what investors expect at each level.
Startup Stage Monthly Burn (2025-2026 Data)
Based on data from Carta’s 2025 analysis of over 12,400 startups and insights from CFO Advisors, startup spending increases significantly as companies move through funding stages.
Early-stage startups spend relatively small amounts, but costs rise quickly as teams grow and operations expand. AI startups, in particular, tend to have higher expenses due to technology and infrastructure needs, and many spend more than they earn in the early stages to support rapid growth.
| Funding Stage | Monthly Burn Rate (Median) | Team Size |
| Pre-seed | $10K–$25K | 2–3 people |
| Seed | $75K–$100K | 4–10 people |
| Series A | ~$250K–$350K | 15–20 people |
| Series B | ~$900K | 50–80+ people |
| AI-focused tech (general) | $100K–$500K+ | Varies |
For AI startups, monthly spending usually falls between $100,000 and $500,000 or more. Most of these companies spend about 1.5 to 2.5 times what they earn in new revenue, and some fast-growing ones spend even more, over 3 times their revenue.
The Burn Multiple: The Key Investor Metric
Burn multiple is a simple way investors measure how efficiently a company is using its money. It shows how much a company spends to generate new revenue.
| Performance Tier | Burn Multiple | Investor Signal |
| Exceptional | <1.0x | Top 10% (elite efficiency) |
| Strong | 1.0x–1.5x | Top 25% (new Series A target) |
| Median (SaaS) | 1.5x–2.0x | Acceptable for most stages |
| Concerning | 2.0x–3.0x | Bottom 25% (fundraising risk) |
| Critical | >3.0x | Bottom 10% (red flag) |
For comparison, most Series A SaaS companies have a burn multiple of around 1.6x. However, some top AI startups are doing better, keeping it below 1.0x by using AI to grow revenue without hiring too many people. A well-known example is Midjourney, which has a small team of about 11 people but generates around $200 million a year.
As startups grow, investors expect them to use money more efficiently. For Series B companies (around $8 million to 15 million in revenue), the target is to spend about 0.8x to 1.2x to generate new revenue. For Series C ($15 million+), this improves further to about 0.5x to 1.0x. At larger scale stages ($20 million+), top companies usually stay in the 1.0x to 1.5x range.
Investors also expect startups to have enough cash to run for 18 to 24 months. For Series A funding, companies are now typically expected to have at least $1.5 million in annual revenue along with this runway.
Why AI Burn Rates Are Structurally Higher
AI companies operate with significantly higher costs than traditional software businesses due to the unique demands of building and running advanced AI systems. Unlike typical SaaS models, they require massive computing power, highly specialized talent, and continuous investment in infrastructure.
These factors make high spending a core part of the business, not just a temporary phase, which is why AI companies tend to have structurally higher burn rates as they scale.
The High Cost of Running AI Models
AI companies spend huge amounts on computing, much more than traditional software companies ever did. For example, training models like ChatGPT can cost millions of dollars, and running them every day is also very expensive.
These systems can cost hundreds of thousands of dollars daily to operate, and each user query adds to that cost. For companies like OpenAI, compute is one of the biggest expenses, expected to reach around $10 billion to 11 billion per year by 2026. In fact, they still spend about $2 for every $1 they earn just on running the models, even before research costs.
Spending on AI infrastructure is also rising quickly. In the first half of 2024 alone, global spending nearly doubled, with most of it going toward powerful GPU-based servers needed to run AI systems.
ALSO READ: AI API Cost Statistics -Enterprise LLM API Cost Surges 140% by Mid-2025
The Talent Premium
AI companies have to pay very high salaries to hire skilled engineers and data scientists. Because competition is so intense, some companies are offering huge pay packages. For example, Meta has reportedly offered up to $100 million in signing bonuses to attract talent from OpenAI.
This strong competition for talent makes it much more expensive for AI startups to hire, which increases their overall spending.
Gross Margin Compression
Traditional SaaS companies usually aim for high profit margins of around 70–80%. AI companies, however, are much lower than this. For example, Anthropic had a margin of about 40% in 2025, below its own expectations, while OpenAI was around 46%.
Neither company is expected to generate positive cash flow until later in the decade. A big reason is that the cost of running AI models (per user query) is not decreasing fast enough, even as their revenue grows.
AI Company Burn Rate: Risks, Failures, and Capital Pressure
High spending in AI is not just a growth strategy; it also brings serious risks. Many startups are burning cash quickly without clear returns, which increases the chances of failure. The data below highlights how burn rates are closely linked to startup risk and survival:
- Around 90% of AI startups fail, compared to about 70% for traditional tech startups.
- About 85% of AI startups shut down within three years, according to industry estimates.
- The biggest reason for failure (42%) is a lack of real market demand.
- Nearly 95% of generative AI pilot projects in companies fail to show clear returns.
- A study by MIT Media Lab found that even after $30 billion to 40 billion in enterprise AI spending, 95% of projects deliver no return on investment.
- A 2025 survey by EY estimated $4.4 billion in losses from AI projects that didn’t meet expectations.
- Startup cash runway has dropped from about 16 months to around 12 months since 2022, while the time between funding rounds has increased significantly.
Overall, this shows that while AI offers huge potential, high burn rates and unclear returns make it a very risky space for many companies.
GenAI Burn Rate: Growth and Projections
Spending across the generative AI (GenAI) industry is expected to keep rising every year. Estimates show total cash burn growing from about $15 billion in 2024 to over $30 billion by 2027. Altogether, leading GenAI companies could spend more than $100 billion by 2027.
Even though companies like OpenAI are growing revenue very quickly, they are still not making profits. This is because the cost of running AI systems grows just as fast or even faster than, their revenue.
| Year | Estimated GenAI Sector Burn |
| 2024 | ~$15 billion |
| 2025 | ~$20 billion |
| 2026 | ~$25 billion |
| 2027 | ~$30+ billion |
Despite this, investors are still heavily backing AI. In 2025, more than half of all venture capital went into AI, and large late-stage deals have become much bigger. The belief is that once companies reach enough scale and dominance, they will eventually become highly profitable, but when that will happen is still uncertain.
Wrapping Up
The AI industry is moving into a stage where efficiency matters as much as growth. Spending will likely keep increasing, especially on computing and infrastructure, but companies will need to focus more on making profits, not just growing fast. Investors are now paying closer attention to how wisely companies use money, not just how quickly they expand.
Even with high failure rates, strong investment in data centers and technology shows confidence in AI’s future. Going forward, the companies that can grow while controlling their spending are the ones most likely to succeed.
