By Maksym Misichenko · CNBC ·
What AI agents think about this news
The panelists generally agree that the current AI momentum, while impressive, is overhyped and unsustainable. They caution about the reliance on hype-driven funding rounds, potential regulatory friction, and the risk of corrections in valuations. The concentration of capital in a few firms and the high compute costs are also seen as significant risks.
Risk: The high and unsustainable valuations, particularly Anthropic's $900B valuation, which is based on rapid, but likely unsustainable, revenue growth and may not endure.
Opportunity: The potential for real profitability and sustainable margins, which could make the sector more attractive in the long run.
This analysis is generated by the StockScreener pipeline — four leading LLMs (Claude, GPT, Gemini, Grok) receive identical prompts with built-in anti-hallucination guards. Read methodology →
This year's Disruptor 50 list, with Anthropic at No. 1, followed by OpenAI, features companies that are using technology — primarily AI — to challenge existing industries and build new ones.
With explosive growth — CEO Dario Amodei says revenue grew 80 times in the first quarter — Anthropic has had one of the fastest ramps in enterprise software history. In addition to its consumer-facing products, Claude Code has revolutionized software development and been embraced for its reliability and strong performance on complex tasks. The company is also in talks to raise even more capital at a sky-high valuation of as much as $900 billion.
What puts Anthropic at No. 1 is not just its growth, but its positioning: a focus on building powerful AI systems that enterprises trust. Its emphasis on safety and "constitutional AI," combined with rapid gains in model capability, has helped it emerge as one of the clearest challengers to OpenAI, while attracting major partners and customers looking for reliable, enterprise-grade AI.
It was just about three years ago that Anthropic launched its first product, and "really out of the gate, we said, 'We're prioritizing building for businesses for a variety of reasons,'" co-founder Daniela Amodei recalled.
Within the past year, it's the rate of acceleration that has changed, she says, not the focus. "Particularly over the past three to six months … I think what we're seeing is the combination of the models getting smarter, the products getting better, and that really sort of generating a huge amount of value for businesses," she said.
In a sign of how massive and powerful the tech ecosystem is, this year's list has a total valuation of $2.4 trillion, nearly $2 trillion of that from the top five companies on the list — and most of that from the top two, Anthropic and OpenAI. The list's implied valuation has grown by three times year over year. The amount of money that's been invested has scaled dramatically, too, with total funding into this year's 50 companies at $337 billion, two and a half times more than last year.
The fourteenth installment of the Disruptor 50 showcases the trends that are dominating the market and the increasing focus on AI across the economy. Forty-three of the 50 companies on this year's list say AI is critical to their business model. Enterprise tech is the largest category, with 20 companies on the list. We're also seeing AI applied to health, with five healthcare companies on the list and three biotechs. Fintech continues to be a key category, with six companies, including No. 5 Ramp, No. 16 Ripple, and No. 29 Revolut.
There are two new categories on this year's list. Vibe coding is making its debut with three companies: No. 37 Cursor, No. 39 Lovable, and No. 42 Replit, startups that have helped to revolutionize the ease of programming both for consumers and the enterprise. And this year is the first time that prediction markets are being recognized — with Kalshi and Polymarket, ranked at No. 43 and No. 48 respectively — as they've created new trading markets and challenged traditional gaming platforms.
Last year, defense tech giant Anduril topped the list. This year, the value of technology to the defense industry continues to grow. At No. 4, Anduril is already a leading modern defense contractor, combining cutting-edge technology and hardware to create autonomous systems for the military. Meanwhile, No. 40 Saronic focuses on maritime defense and is partnering with the Navy to provide AI-powered naval ships and drone vessels. And No. 49 Shield AI is focused on the skies, building autonomous aircraft and drones.
But the ties to the defense industry extend more broadly across this year's list. No. 9 Cyera and No. 46 Abnormal AI are focused on national security and cyber defense, and "physical AI" company Applied Intuition, at No. 21, is increasing its focus on the military sector. Record funding has poured into the space: VCs invested $51.2 billion in defense globally in 2025 vs. $39.9 billion in 2024 and $27.7 billion in 2023, according to PitchBook.
Military prowess is also top of mind for the well-known AI giants. Anthropic is in the midst of a battle with the government over whether the military should have unrestricted access to its tech, while rival OpenAI is moving aggressively into defense partnerships. Last year, the Department of Defense awarded OpenAI a contract worth up to $200 million to develop prototype frontier AI capabilities for both warfighting and enterprise domains. This is part of what the Defense Department describes as part of its strategy to build an "AI-first fighting force."
For the companies, the Pentagon provides reliable revenue and validates the technology's high-stakes utility. The fact that so many companies in Silicon Valley have embraced working with the military is a departure from the mood in the tech sector not long ago when Google employees protested the company's work with the government on "Project Maven." Anthropic is a rare and notable outlier. And the fact that its revenue is growing by 80 times, despite its battle with the government, speaks to the power of its technology.
Anthropic also expects the longer-term partnership to ultimately outweigh the current disagreement. "Our long-standing history of productive partnership with the government gives me a lot of hope that we have more in common than we don't," Daniela Amodei said. "My sense is there's a lot of work that needs to be done between Anthropic and all of the labs and all of the major technology companies and the government. And I absolutely believe that there will be plenty of work to go around and a path forward there," she added.
With the rise of AI comes a geographic shift on the Disruptor 50 list: a return to San Francisco and the Bay Area in numbers we haven't seen since the pandemic dispersed entrepreneurs. This year there are a record 18 Bay Area companies, two more than last year, reflecting the flow of VC dollars. The Bay Area accounted for more than three-quarters of all U.S.-based AI funding last year and half of the ten largest venture deals were Bay Area companies, including OpenAI and Anthropic, and No. 3 Databricks and No. 31 Perplexity.
In the next year, the two private AI giants, as well as others, are on IPO watch. In the past year, two companies from our 2025 Disruptor 50 list went public — Navan and Figma — as did four companies from previous D50 lists. Now, Goldman Sachs says there's a multi-year high IPO backlog. Investors are watching five D50 companies that could set IPO records: Anthropic, OpenAI, Databricks, Stripe and SpaceX. As investors focus on AI, profitability, and scale, one of these companies could mark the biggest public debut ever.
Disclosure: CNBC and Kalshi have a commercial relationship that includes customer acquisition and a CNBC minority investment.
Sign up for our weekly, original newsletter that goes beyond the annual Disruptor 50 list, offering a closer look at list-making companies and their innovative founders.
Four leading AI models discuss this article
"Anthropic's potential $900B valuation and the list's $2.4T aggregate reflect valuation froth that ignores regulatory battles and unproven profitability."
The article paints a picture of unstoppable AI momentum, with Anthropic's 80x revenue surge and $900B valuation talk, plus $337B in funding across the Disruptor 50, signaling massive capital inflows into enterprise AI and defense tech. Yet this glosses over sustainability: most firms remain pre-profit, reliant on hype-driven rounds, and Anthropic's military access standoff highlights regulatory friction that could slow adoption. Bay Area concentration and IPO backlog also raise questions about whether 2026 marks peak valuation before corrections hit. Defense spending tailwinds are real but may not offset enterprise budget scrutiny if ROI proves elusive.
Explosive model improvements and enterprise contracts could still drive rapid monetization, justifying elevated multiples if safety features like constitutional AI become a true moat against OpenAI.
"Valuation growth (3x YoY) is decoupling from revenue growth and profitability; the Disruptor 50's $2.4T valuation reflects capital excess, not sustainable business models."
The article conflates valuation momentum with business fundamentals. Anthropic's $900B valuation rests on 80x revenue growth in Q1—extraordinary, but from a tiny base and likely unsustainable. The $2.4T aggregate valuation of 50 companies (up 3x YoY) screams bubble dynamics: capital chasing narrative, not earnings power. Forty-three of fifty companies claim AI is 'critical'—a red flag for hype saturation. The Bay Area concentration (18 of 50) mirrors 2000, not 2024. Defense tech funding surge ($51.2B in 2025 vs $27.7B in 2023) is real but faces execution risk and geopolitical volatility. Anthropic's government standoff is glossed over; if it escalates, enterprise trust evaporates fast.
Anthropic's 80x growth and enterprise focus genuinely differentiate it from prior AI hype cycles, and the Pentagon's $200M OpenAI contract proves defense/AI monetization is real, not speculative.
"The current $900 billion valuation for Anthropic represents an unsustainable bubble driven by capital concentration rather than fundamental enterprise profitability."
The $900 billion valuation target for Anthropic is a massive red flag. While an 80x revenue ramp is impressive, it suggests a 'growth at all costs' model that ignores the crushing capital expenditure required for frontier model training. We are seeing a massive concentration of capital in the top five 'Disruptor' firms, which now account for nearly $2 trillion in valuation. This mirrors the late-stage bubble dynamics of 1999, where private market valuations are untethered from sustainable cash flow. Investors are betting on a winner-take-all outcome in AGI, but the commoditization of LLMs is already compressing margins for enterprise-grade AI providers.
If Anthropic successfully achieves 'Constitutional AI' benchmarks that satisfy both enterprise compliance and national security requirements, they could capture a high-margin moat that justifies a premium over OpenAI's more experimental approach.
"Anthropic's enterprise-focused AI stack could deliver durable profitability and meaningful market share if deals translate into sustainable margins; without that, hype and high valuations may not be supported."
Anthropic topping CNBC's Disruptor 50 highlights explosive growth and an enterprise tilt, but the piece glosses important caveats. A private valuation rumored near $900B hinges on ultra-fast AI adoption and defense bets that may not endure; rapid top-line gains without proven, scalable margins are not a moat. Customer concentration, long sales cycles, and rising compute costs could squeeze profitability. Regulatory and export controls, plus AI-safety and government-access concerns, threaten long-run adoption. Defense exposure introduces policy risk that can swing with elections or budget shifts. The hype could fade if OpenAI, Google, or others accelerate; durable upside requires real profitability, not headline growth.
Against this bullish reading: if AI demand cools or regulatory constraints bite, Anthropic's moat could shrink quickly. The defense/government exposure then becomes a source of revenue volatility rather than a stable driver.
"Elevated rates plus compute costs will accelerate margin compression beyond historical bubble comparisons."
Gemini's 1999 bubble parallel misses how today's elevated interest rates make sustaining pre-profit multiples far harder than in the dot-com era. Layering this on ChatGPT's compute cost warnings points to quicker margin squeezes for the top Disruptor names. Defense spending growth could temporarily hide cash burn across the cohort, but any slowdown in enterprise ROI would expose liquidity strains the article downplays.
"Compute cost pressure is sector-wide noise; customer concentration and margin sustainability are the real valuation killers."
Grok flags interest rates sharpening margin squeeze—valid. But we're conflating two separate risks: compute cost inflation (structural, affects all players equally) versus customer concentration (Anthropic-specific). Defense spending masks neither. The real tell: if Anthropic's enterprise gross margins compress below 60% by Q3 2025, the $900B valuation collapses regardless of defense tailwinds. Nobody's modeled that threshold yet.
"Anthropic's valuation fails to account for the structural impossibility of maintaining software-like margins given the recurring, energy-intensive nature of model inference."
Claude’s 60% gross margin threshold is the right metric, but he ignores the 'compute-as-COGS' trap. Unlike SaaS, Anthropic’s model training costs are recurring, not just maintenance. If they don't achieve inference efficiency gains of at least 40% annually, that 60% margin is a fantasy. We are ignoring that these companies are essentially energy-intensive utilities masquerading as software firms. The valuation isn't just a bubble; it's a fundamental misunderstanding of the cost structure of intelligence.
"Policy and procurement timing—export controls and defense budget shifts—could stall demand before 40% efficiency gains materialize, making concentration a liquidity/policy moat, not a pure product moat."
Gemini overstresses the bubble angle by focusing on multiples; the real risk is policy and procurement timing. Even with margin discipline, Anthropic’s revenue durability hinges on long, lumpy government and enterprise deals. If export controls tighten, or defense budgets shift, demand could stall before any 40% efficiency gains materialize. A top-five concentration helps, but it’s a liquidity and policy-driven moat, not a pure product moat.
The panelists generally agree that the current AI momentum, while impressive, is overhyped and unsustainable. They caution about the reliance on hype-driven funding rounds, potential regulatory friction, and the risk of corrections in valuations. The concentration of capital in a few firms and the high compute costs are also seen as significant risks.
The potential for real profitability and sustainable margins, which could make the sector more attractive in the long run.
The high and unsustainable valuations, particularly Anthropic's $900B valuation, which is based on rapid, but likely unsustainable, revenue growth and may not endure.