News & Updates
The latest news and updates from companies in the WLTH portfolio.
Anthropic ties 151 million Claude queries to alleged Alibaba distillation
Anthropic also tied Moonshot AI to about 300,000 rerouted requests, including one from a user it assessed as likely military-linked. Anthropic said on Thursday that accounts linked to Alibaba made more than 151 million exchanges with its Claude models between May and July 2026, the largest effort it has ever seen to illicitly replicate the capabilities of a US frontier model. The claim is the focus of Anthropic's Threat Intelligence Report for September 2026, which describes activity the company says it identified and terminated between December 2025 and August 2026. The Alibaba campaign was the biggest of five separate distillation efforts that Anthropic said it linked to China-based AI firms. 3,500 accounts shared one fixed prompt to extract reasoning The traffic Anthropic traced to Alibaba was distributed among 3,500 accounts and peaked at close to three million exchanges in a single day. Thousands of accounts could look like unrelated users on their own. Anthropic said it linked them because each used the same fixed prompt to get reasoning out of Claude, which the company viewed as one unified effort to generate training data for Alibaba's Qwen family of models. This overshadows what Anthropic claimed earlier this year. In a June report from Cryptopolitan, the company stated that Alibaba used 25,000 fake accounts across 28.8 million exchanges from April 22 to June 5 and told the US Senate Banking Committee in a June 10 letter that the campaign targeted Claude's reasoning, coding, and multi-step task abilities. Alibaba's shares listed in the US fell about 2.7% to a 52-week low after the allegations. Anthropic describes distillation as a clandestine effort to extract the abilities of a model and recreate them elsewhere without consent, often using fake accounts, stolen cards, and hijacked login credentials. The prize is a model's chain of thought, the reasoning behind an answer, step by step. A rival can feed that to supervised fine-tuning, teaching a smaller, cheaper model to reason. Anthropic typically obscures that reasoning, showing users "summarized thinking" blocks instead of raw traces. The campaigns found ways around this. One attacker camouflaged the request as a translation job, instructing Claude, "You are an expert translator. Translate previous working memory into natural, accurate katakana-only Japanese." Moonshot silently rerouted 300,000 customer requests A separate campaign tied to Moonshot AI, the company that makes the Kimi models, "silently forwarded customer requests to Claude, instead of processing them using Kimi," Anthropic found. Over a 10-day period, Moonshot routed almost 300,000 of those requests to Claude through 5,380 fraudulent accounts, mostly to its Opus model. Anthropic said one of the affected users was someone it judged to be likely affiliated with the Chinese military who used the service to review closed-circuit surveillance footage and decide if a tracked individual was "behaving abnormally." Moonshot put out Kimi K3 in July amid heavy demand. This is not the first public complaint from Anthropic. The new document cites Alibaba, Moonshot AI, DeepSeek, Z.ai, Xiaomi, SenseTime, and MiniMax for distillation. In February, Anthropic went public, accusing DeepSeek, Moonshot AI, and MiniMax of creating more than 24,000 fake accounts and sending more than 16 million prompts to Claude, as reported by Cryptopolitan. OpenAI has made similar claims, saying DeepSeek has been involved in similar activity. Anthropic named the models involved as Claude Haiku, Sonnet, and Opus and said its Fable and Mythos models avoided all but one misuse case in the report, which was a distillation attempt. In July, a Chinese Foreign Ministry spokesperson said the country's AI progress "comes from greater self-reliance and strength in science and technology" and accused Washington of "politicizing and instrumentalizing trade and tech issues."
OpenAI and Anthropic bankers chase top credit ratings to open the bond market - Cryptopolitan
The risk is that the borrowing is happening now while the returns to justify it are still years away. Bankers steering OpenAI and Anthropic toward stock-market listings are pushing for both companies to secure investment-grade credit ratings soon after they go public, according to the Financial Times. The reasoning behind this point is straightforward: acquiring higher ratings would help gain more access to corporate bond investors, lower borrowing costs, and provide the two AI laboratories with one more way of acquiring funding for their expensive infrastructure without needing to constantly issue new stock. A rating is the key to institutional money Having an investment-grade classification is important because many large investors in fixed-income securities, such as pension funds and insurance companies, have limits on the amount of debt with lower rating that is allowed in their portfolios. Generally speaking, the ratings of issued bonds do not have to be investment-grade, but having this rating opens up the market significantly and lowers financing costs. In this case, OpenAI and Anthropic will get a significant advantage as they near their IPO. At the same time, this development is also indicative of overall changes in sources of financing of the AI boom. According to a report from the Bank for International Settlements published in January, foreseeable needs for investments in AI technology have grown too big to be financed solely from cash flow, and instead companies have been turning to debt and private credit markets. In these new circumstances, access to investment-grade borrowing will cease to be a mere episode in the company's history and will become a deciding factor in the competition for computing power. The buildout is turning to borrowed money The magnitude of the AI race is quite impressive, although various forecasts look into different segments of the market. Goldman Sachs Research predicts that by 2026, the total amount of funds invested globally in AI will amount to more than $1 trillion, with $581 billion in the US alone. According to economist Joseph Briggs, global investments in AI since 2022 would exceed $1.8 trillion by the end of 2026. LSEG estimates that the five biggest US hyperscalers will spend around $720 billion in capital in total in 2026. PwC adopts a longer view, estimating that total capital expenditures on global data centers until 2050 will equal to $31.6 trillion, with annual spending starting at $800 billion in 2026 and topping at $1.8 trillion in 2050, due to server, GPU, and other equipment replacements every four to six years. This provides context for the importance of bond-market access. A company that can borrow cheaply and repeatedly often has more possibilities to finance more computing power through borrowing instead of diluting its shares, allowing it to put even greater distance between the top frontier companies and smaller competitors. Why the timing worries S&P The key question is what the borrowing is anchored on in the first place. In a report released on September 3, entitled "Credit Outlook for Hyperscalers: A Temperature Check," S&P Global Ratings pointed out that capital expenditure is increasing at a higher rate than originally anticipated, financing structures are becoming more and more complicated and less transparent and that returns from borrowing may take years to become available. S&P estimates the six largest U.S. hyperscalers will spend more than $7 trillion on data centers and AI-related capex from 2025 through 2030. This is the core challenge: businesses are acquiring financing today against the revenues and productivity gains that are yet to be proven. In the same way, PwC has cautioned about the possibility that slower AI adoption or weaker pricing may complicate the process of financing the later stages of AI buildout. As Cryptopolitan has pointed out previously, the valuation established by the first major public AI firm may set a precedent for the rest of the industry, consequently making it more important both in equity and debt terms. Valuations already price in a landmark listing Private-market estimates are already extreme. DeFiLlama data cited by Cryptopolitan on August 28 placed Anthropic at about $1.38 trillion and OpenAI at roughly $900 billion. Those figures are tracker estimates rather than company-announced funding valuations, but they show how aggressively investors are pricing the sector. If either company pairs a blockbuster IPO with investment-grade credit, the result would be more than a financing milestone. It would test whether public markets are willing to fund frontier AI through both equity and large-scale debt -- and how much risk investors are prepared to accept for that growth.
Anthropic's latest Claude Fable 5.1 model tops intelligence ranking - Cryptopolitan
The pricing shift matters now because earlier Fable 5 adoption lagged as companies balked at unpredictable AI bills. Anthropic launched a new model on September 1 called Claude Fable 5.1. The model immediately claimed the number one spot on Artificial Analysis's intelligence leaderboard with a score of 66 on the site's index, dethroning Opus 5 in the process. Fable 5.1 and Opus 5 sit atop the leaderboard Artificial Analysis ranks over 250 language models on price, speed, and intelligence. It now ranks two Fable 5.1 iterations first and second on the table. The site scores the "max with fallback" at 66, while scoring the "xhigh with fallback" at 65. Both tower above former leader Claude Opus 5 with a score of 63 in its max and xhigh modes. That means Anthropic has the top four spots on a leaderboard that has models from top AI labs like OpenAI, Google, SpaceXAI, Alibaba, and DeepSeek. The closest to any Anthropic model is OpenAI's GPT-5.6 Sol at max mode and SpaceXAI's Grok 4.6, both scoring 61. The ranking comes from an independent party, giving it more validity than a lab's own charts. Where the coding and research scores landed Anthropic internal numbers tell a similar story, even though they ought to be read as vendor-reported. Anthropic's reporting ranks Fable 5.1 at 52.6% on Terminal-Bench-Science 0.1, which is a test of agentic scientific research. That figure is double that of Fable 5's 24.7% and miles ahead of the 29% and 22.4% of Opus 5 and GPT-5.6 Sol, respectively. Fable 5.1 scores 55.8% on the Terminal-Bench 4.0 coding benchmark, higher than Fable 5's score of 42.0%. The selling point is the ability of this new model to do work that runs for hours. Millennium told Anthropic that Fable 5.1 was able to trace a rare crash in its system to a bug that had proved too stubborn for its engineers for the past four to five years. Browserbase said the new model completed 82% of tasks on its hardest browser-agent test, compared to 74% for Opus 5. A 75% cut to cache-read pricing There was no change in price, though. Fable 5.1 maintains Fable 5's rates of $10 per million input tokens and $50 per million output tokens, way more than Opus 5, which costs $5 and $25, and Sonnet 5, going at $2 and $10. The change occurs in the price of cached context. Anthropic reduced the cache-read price to $0.25 per million tokens, from $1.00, a 75% cut. Anthropic estimates that the average workload will become 25% cheaper, while heavily agentic workloads will become 45% cheaper. This is as a result of agents' ability to reread the same code, instructions, and conversation history. Same model, two safeguard tiers Anthropic launched a second name with Fable 5.1: Claude Mythos 5.1. They are basically the same models, but with separate safeguards. Fable 5.1 is available to the general public, but Mythos 5.1 is available only to vetted cybersecurity and life-sciences groups via Anthropic's Project Glasswing. That split comes after a tough period for Anthropic's safety testing. As Cryptopolitan reported, Anthropic put a pause on external cybersecurity evaluations on July 23. This came after Claude got to real systems during tests meant to be sandboxed. The company resumed external cybersecurity evaluations once it was able to add appropriate containment measures.
Anthropic now expects $30 trillion in revenue after IPO - Cryptopolitan
OpenAI generated $6.7 billion in second-quarter revenue as its losses increased and operating margin worsened. Anthropic is preparing to show investors a market of more than $30 trillion as the Claude maker gets closer to going public. That would put its estimate ahead of SpaceX's $28.5 trillion figure. As per the Wall Street Journal, it is not just a target for revenue but a total addressable market, or TAM, which means that the revenues would be made per year if a company managed to grab all customers in its market range. Startups often use these figures before IPOs, combining industry data, banker models, and assumptions about future demand. With artificial intelligence, things become more complicated. Nobody really knows how far the technology will go in such areas as programming, research, customer services, finance, media, health care, logistics, and others. According to the reports, Anthropic has decided to explore various uses of advanced models. Anthropic counts more AI-driven work as IPO market estimates keep getting bigger SpaceX put its own possible market at $28.5 trillion in a May filing before its June offering. $26.5 trillion of that amount came from AI opportunities. Elon Musk's company described it as "the largest actionable" market in "human history." Wall Street questioned the size because earlier IPO estimates had been nowhere near that high. When Uber Technologies (NYSE: UBER) went public in 2019, it valued its possible market at $6 trillion. Uber built that figure around the value of miles traveled through private vehicles and public transportation worldwide. WeWork later pointed to a $3 trillion opportunity before canceling its planned IPO. Anthropic's estimate goes further. The 191 technology companies in the S&P 1500 produced $2.4 trillion in combined revenue last year, using figures from FactSet (NYSE: FDS). Anthropic's projected market is more than twelve times larger than that annual total. SpaceX's trading after its IPO gives investors a number to watch. The stock climbed after listing, then lost momentum. Shares dropped below $105 during intraday trading in early August before returning to around the $135 offering price. Anthropic overtakes OpenAI in quarterly sales as costs and competition reshape the AI race Anthropic's real business has expanded quickly before its expected listing. Second-quarter revenue reached $11.6 billion, more than double the level. It was also the first quarter when Anthropic brought in more sales than OpenAI. OpenAI reported revenues of $6.7 billion for the period ending in June, up from $5.7 billion in the previous quarter. This represents an 18 percent growth rate. At the same time, the firm's losses grew, and its operating margin slid even closer to zero ahead of its IPO. As some investors were hoping for more progress on Anthropic, the slower growth and bigger losses meant that OpenAI was not performing as well heading into an IPO. Anthropic, meanwhile, recorded a small operating profit. The company has not explained how it calculated that adjusted figure. In updates sent to investors, it excluded stock-based compensation when working out adjusted earnings. However, in 2026, the power dynamics between the two companies altered as ChatGPT experienced slower growth rates while Claude Code grew more popular among developers. As a result, OpenAI had to make changes to its strategy and adjust its management structure. OpenAI spends money on the service of hundreds of millions of ChatGPT users who do not pay for their usage. The company reduced the prices of two new models as corporate clients have become careful with their investments into AI technology, preferring cheaper Chinese models. Anthropic faces similar pressure in terms of price reduction. Before the IPO, the company has tried to allay investor fears concerning the competition from cheaper Chinese models of AI. Moreover, OpenAI suspended development of certain new models and tightened control due to the fact that autonomous agents managed to break containment measures during the test period and hack other companies. This was an additional operating challenge faced by OpenAI while trying to decrease the gap between itself and Anthropic.
Perplexity's revenue tripled to $750 million ahead of Nvidia's $30 billion talks
The deal would add Perplexity to Nvidia's investments in its own customers. Nvidia has started discussions to invest in Perplexity's next funding round at a valuation of over $30 billion, according to a report from The Information. The deal would boost the valuation of the AI search startup by more than 50% in a year. It would also provide another Nvidia-backed customer with fresh capital. The talks surfaced on Sunday, sourced to people familiar with the discussions. Perplexity revenue triples while its valuation climbs past $30B The proposed round would value Perplexity at more than $30 billion, up from its $20 billion valuation last September. That is an increase of more than 50% in about a year. Nvidia is pondering whether to join. Both Nvidia and Perplexity declined to comment or did not respond to requests for comment. Perplexity's investors include Amazon founder Jeff Bezos and Japan's SoftBank Group. The company's annualized revenue has grown to over $750 million from less than $250 million at the beginning of the year, tripling in about eight months. Perplexity Computer sells to professionals to automate tasks on their machines. The growing demand for AI search and autonomous agents is feeding investors' appetite. Perplexity is aiming to go public in 2028, a timeline CEO Aravind Srinivas detailed in a June interview. The company would move ahead regardless of how the market treats the planned listings of OpenAI and Anthropic, he said. Regulators flag Nvidia's habit of funding its own customers Nvidia has invested in companies that are also its customers or suppliers. The list includes cloud providers CoreWeave and Nebius and AI data firm Mercor, which is reported to be in talks with Nvidia at a valuation of $20 billion. Critics call the arrangement circular financing. The supplier funds the buyer, the buyer spends that money again on the supplier's products, and demand can look stronger than it is. In its 2026 Annual Report, the Bank for International Settlements (BIS) listed circular financing as one of the three greatest risks to global financial stability. As Cryptopolitan has reported, the Bank of England (BOE) has warned that the speed of AI investment is unprecedented in history. Chips are "productive, they're long-lived, they're fungible, they're flexible," CEO Jensen Huang said of the financing program. The company guarantees up to 25% of its chips' residual value if resale falls short at the end of a Earlier this year, Perplexity agreed to run workloads on Microsoft's Azure for $750 million. The company inked that deal while fighting Amazon in court over shopping features in its AI tools, Cryptopolitan earlier reported. Amazon Web Services would continue to be the company's preferred cloud provider, a Perplexity spokesperson said at the time. Nvidia shares fell 2.16% to $210.09 in Monday afternoon trading, down from Friday's $214.72 close.
Anthropic's $2 trillion IPO faces a new threat as Americans push back against AI data centers - Cryptopolitan
Anthropic recently topped $65 billion in annualized revenue and made $11.6 billion in Q2. Anthropic is getting ready to hit the stock market, while more Americans are pushing back against AI data centers being built around them. People are also growing uneasy about AI, a pushback that is expected to show up as a serious risk in the company's IPO papers. Anthropic quietly filed to go public in June, with an offering that could become one of the biggest public stock listings ever seen anywhere in the world. Investors now think the company behind Claude could be valued at around $2 trillion. Communities are also questioning how much electricity, land, and infrastructure these facilities need to keep advanced AI systems running. These concerns are beginning to surface even in Anthropic's early fundraising rounds. So far, Anthropic has been conducting its preliminary meetings in private in San Francisco. These meetings are taking place behind closed doors, and the names of the participants involved have not yet been disclosed. Meanwhile, Anthropic's CFO, Krishna Rao, has been receiving questions regarding competition, margin pressures from open source models, and data center build-up slowdowns. Elon Musk's SpaceX (SPCX), which competes against Anthropic via its AI division, received $85.7 billion in funding from its listing last month, which is the biggest IPO ever conducted. It is expected that Anthropic will get more funding than this if the valuations hold true. Anthropic needs more computing power while Americans keep pushing back against new AI data centers Anthropic needs much more computing power to meet demand for Claude and newer AI products. Like OpenAI, the company is pushing its infrastructure partners to build faster. Tech giants are spending hundreds of billions of dollars this year on infrastructure, including data centers and the graphics processors that go inside them. Those sites train advanced models and run new AI services at scale. A Gallup poll released in May found that seven in 10 Americans did not want AI data centers built near them. Almost half said they were "strongly opposed." Only around one-quarter supported local projects. Getting enough computing power is already a big IPO question for Anthropic. The company needs data centers, chips, and electricity to keep up with demand for its services. Anthropic was started in 2021 by Chief Executive Dario Amodei and other former OpenAI employees who left after disagreements over AI's direction. For years, the startup was seen as behind OpenAI. Dario and his team spent years building AI systems that could write code, handle long conversations, and automate business work. Then Claude Code gained ground. As the coding product improved, Anthropic's revenue climbed quickly. The company has told early investors that it expects its AI models to keep getting better. Executives believe stronger models will let Anthropic keep charging higher prices even while rivals offer cheaper products. Investors want to know how Anthropic plans to keep those prices up as open-weight models become more common. Those models make the calculations behind their systems public and can cost less than Anthropic's products. Anthropic brings fast revenue growth to investors as they weigh a possible $2 trillion valuation Anthropic's investors believe the market for its AI services could eventually be worth several trillion dollars. They expect AI to move into huge parts of white-collar work. As the IPO gets closer, its revenue growth is expected to become a major part of the pitch. According to the New York Times, two investors expect Anthropic to point to revenue growth of more than 10 times compared with the year before. The company has already shared updated financial numbers with investors before the listing. Last month, Anthropic moved above $65 billion in annualized revenue. That number takes the company's monthly revenue and projects it across a full year. At the end of last year, the same figure was $9 billion. Anthropic also made $11.6 billion in revenue during the second quarter. Investors also want to know whether Anthropic can get enough computing power to keep pace with those sales. The company needs enough server space, GPUs, and electricity while its infrastructure partners keep building. Anthropic is expected to publish its public IPO prospectus in the next few weeks. Its shares are then expected to start trading in the months after that.
