Newshot Anthropic Says Its AI Model Cut AMD Chip Switch to Three Days. Why it Could be a Challenge to Nvidia
Anthropic Says Its AI Model Cut AMD Chip Switch to Three Days. Why it Could be a Challenge to Nvidia
Anthropic said its own artificial intelligence model helped get Claude running on AMD chips in three days, accelerating work that previously required months.
The claim, presented to investors ahead of Anthropic's proposed initial public offering, suggests AI coding tools could reduce the engineering burden that makes switching chip suppliers difficult. However, the three-day result covers a specific workload, rather than a complete migration of Claude's training and operations.
Tom Brown, Anthropic's co-founder and chief compute officer, said the company used its Mythos model to handle the transition after beginning work with AMD GPUs this year. According to the presentation, Mythos wrote or ported nearly all the necessary kernels, the specialized software routines that execute calculations on chips. "It got Claude running on AMD in just three days with basically no human intervention," Brown said in the investor presentation.
The approximately three-day timeline refers to bringing up prompt processing for one Anthropic model on AMD MI355X GPUs at a useful performance level. Prompt processing is the stage in which a model reads and processes input before generating its response.
The reported acceleration addresses a significant obstacle for Nvidia's competitors. CUDA provides the software infrastructure developers use to put Nvidia GPUs to work. Companies that have built their systems around that ecosystem can face substantial engineering costs when adapting their software to another supplier's hardware.
If AI can perform more of that adaptation, customers could gain flexibility to choose chips based on availability, performance and cost. Anthropic says it has spent years building that flexibility. Its IPO materials describe a computing architecture spanning four chip families: Nvidia GPUs, AMD GPUs, Amazon Web Services Trainium chips and Google's tensor processing units.
The company treats capacity across its partners as a shared pool, allocating computing resources to research, model development, customer services and internal AI use. Its investor fact sheet identifies both its multiple-chip architecture and research breakthroughs as foundations of that approach.
The strategy could become increasingly important as demand for AI computing grows. Different workloads have different requirements, and access to multiple suppliers gives Anthropic more options when adding capacity or deciding where to run particular tasks.
Anthropic also describes research advances as "compute multipliers," improvements intended to extract more capability from existing hardware. The company's longer-term ambition includes having Claude help discover further efficiency gains. That remains an objective, rather than proof of autonomous recursive self-improvement.
Alongside those efforts, Anthropic is pursuing its Anthropic Direct program, leasing chips and operating them in purpose-built data centers to lower ownership costs. Its relationship with AMD already extends beyond software experiments.
AMD announced in July that Anthropic would deploy up to two gigawatts of Instinct MI450-series GPUs in Helios systems, with the first gigawatt deployment beginning in the first half of 2027. AMD also committed to a strategic equity investment of up to $5 billion.
- AMD Stock: What Does Its $2.5 Billion Anthropic Bet Buy? Trefis Team Thu, November 12, 2026 at 4:02 AM PST 4 min read AMD +4.15% ANTH.PVT NVDA -4.07% GOOG AVGO Image from Pixabay More on the Anthropic IPO: our Anthropic IPO storyline page collects our pieces on it. Maybe Claude turned your messy meeting notes into a to-do list this morning. Some of that work runs on AMD chips: on its Nov 3 call, AMD said one Anthropic engineer used Claude to get Anthropic's latest model running on AMD racks in a single weekend. Everyone knows AMD's week. The stock closed Wednesday at $552.90, up 7.3% on the day and 158% this year (more in Buy, Sell Or Hold AMD Stock At $553?). Anthropic's Nov 10 filing shows AMD buying $2.5 billion of its stock at the IPO price. Fewer know why. AMD's money follows its chips. Under a July supply deal, Anthropic expects to spend more than $20 billion on AMD-based Helios racks if it takes its full first tranche, starting in 2027. AMD in turn pledged up to $5.0 billion of equity, paid in as its GPUs are delivered and deployed. The IPO check is the first half. This 4-to-1 deal arrives with AMD up 158% this year, and in a seven-figure family portfolio that run may have pushed it past its planned weight. By how much? A free 15-minute call with a Trefis partner firm can measure it. What Does AMD Get For $2.5 Billion? AMD and Anthropic Number Shares at $70-$78 35.7M-32.1M (33.8M at $74) Share of Anthropic, fully diluted About 0.2% AMD's equity pledge Up to $5.0B; $2.5B left after the IPO Anthropic's expected AMD rack spend More than $20B, plus an option on about as much again Source: Anthropic S-1/A, Nov 10; counts at $70 and $78 are our arithmetic. Later AMD purchases use Anthropic's 30-day average price. AMD pays the IPO price, so its stake is worth $2.5 billion anywhere in the range; only the share count moves. It is the third-largest check in the $26.0 billion placement, after $10.0 billion each from Google and Nvidia. For its size, AMD's check (0.28% of its market value) matches Google's and tops Nvidia's (0.24%). How Big Is It Next To AMD's Sales? AMD Number Revenue, quarter to Sept. 26 $13.4B Data center sales, same quarter $8.5B, +97% from a year earlier Market value, Wednesday's close $907B Sources: Morningstar (revenue); AMD's Nov 3 earnings call; Trefis prices through Nov 11. The stake is small for AMD: a 1% move in its stock is $9.1 billion, over three times the check. The order is not. More than $20 billion is over two quarters of AMD's latest data center sales, though not all of it lands at AMD; racks come through approved manufacturers or service providers. Has AMD Paid For Its Pain? 5 years to Nov 11 Return a year Worst drop Gain per pain AMD +30.5% −65.4% 0.85 Nvidia +41.9% −66.3% 1.41 Broadcom +46.7% −41.1% 2.32 S&P 500 (SPY) +11.9% −24.5% 0.87 Gain per pain: yearly return above T-bills (3.8%), divided by the typical size of down months. Above 1 is good. AMD took a drop like Nvidia's for less gain. Source: Trefis, adjusted prices through Nov 11. What Should You Do This Week? Size it from the worst drop. With a 1-3% loss limit per stock, a repeat of the 65% fall means holding AMD at about 1.5-4.6% of everything you own, index funds included. Watch dated triggers. About Nov 18: Anthropic's IPO price. AMD's next report: revenue against its guide of about $17 billion. 2027: the first GPU delivery milestones. The last $2.5 billion opens only after milestone credits pass the $2.5 billion AMD pays at the IPO. What it replaces: buying AMD on an IPO headline. The risk: the filing ties Anthropic's AMD purchases to pricing, financing and deployment terms. If those slip, so does the order. Why Does AMD Invest In Its Own Customer? Call it customer financing paid in shares: the more AMD hardware Anthropic deploys, the more Anthropic stock AMD can be asked to buy. It buys AMD a bigger place in a four-chip mix, with some of AMD's own money attached. What Should You Ask Next? Anthropic IPO: Buy, Sell Or Hold Nvidia Stock? Anthropic IPO: What It Means For Amazon, Google And Nvidia Stock Want Income To Go With Your Growth Stocks? Growth stocks like AMD pay off mostly when you sell. Dividend stocks pay you along the way. The Trefis Income Solution holds 50 companies chosen to keep paying dividends, each with at least 10 straight years of dividends, yielding about 3x the S&P 500. See if it fits your plan. For questions about this article or underlying research, email [email protected]. Terms and Privacy Policy Your Privacy Choices CA Privacy Notice More Info
AMD Stock: What Does Its $2.5 Billion Anthropic Bet Buy?
More on the Anthropic IPO: our Anthropic IPO storyline page collects our pieces on it.
Maybe Claude turned your messy meeting notes into a to-do list this morning. Some of that work runs on AMD chips: on its Nov 3 call, AMD said one Anthropic engineer used Claude to get Anthropic's latest model running on AMD racks in a single weekend.
Everyone knows AMD's week. The stock closed Wednesday at $552.90, up 7.3% on the day and 158% this year (more in Buy, Sell Or Hold AMD Stock At $553?). Anthropic's Nov 10 filing shows AMD buying $2.5 billion of its stock at the IPO price.
Fewer know why. AMD's money follows its chips. Under a July supply deal, Anthropic expects to spend more than $20 billion on AMD-based Helios racks if it takes its full first tranche, starting in 2027. AMD in turn pledged up to $5.0 billion of equity, paid in as its GPUs are delivered and deployed. The IPO check is the first half.
This 4-to-1 deal arrives with AMD up 158% this year, and in a seven-figure family portfolio that run may have pushed it past its planned weight. By how much? A free 15-minute call with a Trefis partner firm can measure it.
What Does AMD Get For $2.5 Billion?
| AMD and Anthropic | Number |
|---|---|
| Shares at $70-$78 | 35.7M-32.1M (33.8M at $74) |
| Share of Anthropic , fully diluted | About 0.2% |
| AMD's equity pledge | Up to $5.0B; $2.5B left after the IPO |
| Anthropic's expected AMD rack spend | More than $20B, plus an option on about as much again |
AMD and Anthropic
Shares at $70-$78
35.7M-32.1M (33.8M at $74)
Share of Anthropic, fully diluted
AMD's equity pledge
Up to $5.0B; $2.5B left after the IPO
Anthropic's expected AMD rack spend
More than $20B, plus an option on about as much again
Source: Anthropic S-1/A, Nov 10; counts at $70 and $78 are our arithmetic. Later AMD purchases use Anthropic's 30-day average price.
AMD pays the IPO price, so its stake is worth $2.5 billion anywhere in the range; only the share count moves. It is the third-largest check in the $26.0 billion placement, after $10.0 billion each from Google and Nvidia. For its size, AMD's check (0.28% of its market value) matches Google's and tops Nvidia's (0.24%).
How Big Is It Next To AMD's Sales?
| AMD | Number |
|---|---|
| Revenue , quarter to Sept. 26 | $13.4B |
| Data center sales , same quarter | $8.5B, +97% from a year earlier |
| Market value , Wednesday's close | $907B |
Revenue, quarter to Sept. 26
Data center sales, same quarter
$8.5B, +97% from a year earlier
Market value, Wednesday's close
Sources: Morningstar (revenue); AMD's Nov 3 earnings call; Trefis prices through Nov 11.
The stake is small for AMD: a 1% move in its stock is $9.1 billion, over three times the check. The order is not. More than $20 billion is over two quarters of AMD's latest data center sales, though not all of it lands at AMD; racks come through approved manufacturers or service providers.
Has AMD Paid For Its Pain?
| 5 years to Nov 11 | Return a year | Worst drop | Gain per pain |
|---|---|---|---|
| AMD | +30.5% | −65.4% | 0.85 |
| Nvidia | +41.9% | −66.3% | 1.41 |
| Broadcom | +46.7% | −41.1% | 2.32 |
| S&P 500 (SPY) | +11.9% | −24.5% | 0.87 |
5 years to Nov 11
Return a year
Gain per pain
S&P 500 (SPY)
Gain per pain: yearly return above T-bills (3.8%), divided by the typical size of down months. Above 1 is good. AMD took a drop like Nvidia's for less gain. Source: Trefis, adjusted prices through Nov 11.
What Should You Do This Week?
Size it from the worst drop. With a 1-3% loss limit per stock, a repeat of the 65% fall means holding AMD at about 1.5-4.6% of everything you own, index funds included.
Watch dated triggers. About Nov 18: Anthropic's IPO price. AMD's next report: revenue against its guide of about $17 billion. 2027: the first GPU delivery milestones. The last $2.5 billion opens only after milestone credits pass the $2.5 billion AMD pays at the IPO.
What it replaces: buying AMD on an IPO headline. The risk: the filing ties Anthropic's AMD purchases to pricing, financing and deployment terms. If those slip, so does the order.
Why Does AMD Invest In Its Own Customer?
Call it customer financing paid in shares: the more AMD hardware Anthropic deploys, the more Anthropic stock AMD can be asked to buy. It buys AMD a bigger place in a four-chip mix, with some of AMD's own money attached.
What Should You Ask Next?
- Anthropic IPO: Buy, Sell Or Hold Nvidia Stock?
- Anthropic IPO: What It Means For Amazon, Google And Nvidia Stock
Want Income To Go With Your Growth Stocks?
Growth stocks like AMD pay off mostly when you sell. Dividend stocks pay you along the way. The Trefis Income Solution holds 50 companies chosen to keep paying dividends, each with at least 10 straight years of dividends, yielding about 3x the S&P 500. See if it fits your plan.
For questions about this article or underlying research, email [email protected].
- Anthropic IPO: Buy, Sell Or Hold Nvidia Stock? Trefis Team Thu, November 12, 2026 at 4:02 AM PST 4 min read ANTH.PVT NVDA -4.07% AMD +4.15% AMZN AVGO Image from Pixabay More on the Anthropic IPO: our Anthropic IPO storyline page collects our pieces on it. Tonight you might ask Claude to fix a broken formula in the family budget sheet. The answer arrives in seconds, worked out on a chip you will never see. Which chip matters to anyone who owns Nvidia. Anthropic's IPO filing says it runs Claude on four chip families: Amazon's Trainium, Google's TPUs, and GPUs from Nvidia and AMD. Everyone knows this week's Nvidia story: a Wednesday close of $173.73, 27.4% below its 52-week closing high. That evening, The Wall Street Journal reported the board was preparing to oust CEO Jensen Huang; shares fell about 8% after hours, and no decision had been announced by early Thursday (see Nvidia Stock At $174: Sell After An 18% Drop?). Fewer have read where Anthropic's compute money goes. Its filing ties about $254 billion of named deals to Nvidia GPUs, and $155.4 billion of it sits in deals that can be ended on short notice by early 2027. Its Google and Amazon deals cannot. Our answer: hold Nvidia, sized from its worst drop. That $155 billion is the swing money. In a seven-figure portfolio, index funds hold Nvidia, Alphabet and Amazon side by side, so where it lands moves several holdings. Which way does yours tilt? A free 15-minute call with a Trefis partner firm can map it. Where Does Anthropic's Compute Money Go? Chips Named deals Easy to exit? Google TPUs $282.5B (Google Cloud $121.3B, Broadcom leases $161.2B) No More TPUs, 2028-2031 $125B-$150B, estimated Not yet fixed Amazon AWS $110.0B or more No Nvidia GPUs About $254B (SpaceXAI up to $120.5B, Microsoft $47.2B+, Nscale up to $44.6B, Lambda $41.3B) $155.4B of it, yes AMD GPUs More than $20B No Source: Anthropic S-1/A, Nov 10; sums are ours. SpaceXAI deals become cancellable on 90 days' notice from early 2027; a $34.9B Lambda deal is terminable now (Anthropic doesn't plan to end it). Google's TPUs and Amazon's AWS: $392.5 billion of fixed money. Our Nov 2 piece, Anthropic Has Promised $417 Billion For Computing, counted contracts signed January to May in the first filing; the amended one puts total commitments near $900.9 billion. Is That A Problem For Nvidia's Sales? Nvidia vs Anthropic Number Nvidia data center sales, quarter to July 26 $89B, +18% on the quarter Anthropic contracted payments due in 12 months after Sept. 30, all suppliers $70.7B Nvidia's Anthropic purchase at the IPO $10.0B, about 0.75% of Anthropic, fully diluted Anthropic's contracted 12-month payments, every supplier included, are less than Nvidia's data center sales in a single quarter. Sources: Nvidia's Aug 26 call; Anthropic S-1/A. Asked in August about labs building their own chips, Huang said he expects them to keep using Nvidia for much of their computing. Has Nvidia Paid For Its Pain? 5 years to Nov 11 Return a year Worst drop Gain per pain Nvidia +41.9% −66.3% 1.41 Broadcom +46.7% −41.1% 2.32 Alphabet +15.2% −44.3% 0.61 S&P 500 (SPY) +11.9% −24.5% 0.87 Gain per pain: yearly return above T-bills (3.8%), divided by the typical size of down months. Above 1 is good. Source: Trefis, adjusted prices through Nov 11. So Buy, Sell Or Hold Nvidia This Week? Size it from the worst drop. With a 1-3% loss limit per stock, a repeat of the 66% fall means holding Nvidia at about 1.5-4.5% of everything you own, index funds included. Watch dated triggers. Nov 17: Nvidia's revenue against its $108 billion guide. About Nov 18: Anthropic's IPO price. Early 2027: the SpaceXAI deals become cancellable. And any board announcement on Huang. What it replaces: selling Nvidia on an IPO headline. The risk: other big labs copy the four-chip setup, and more flexible money moves to custom chips. Why Would Nvidia Buy Into A Customer That Shops Around? Anthropic's four-chip setup cuts its dependence on any one vendor. Nvidia's answer: own part of the customer. Its $10 billion is about 0.24% of Nvidia's $4.19 trillion market value, a small price for a seat at the table. What Should You Ask Next? Anthropic IPO: What It Means For Amazon, Google And Nvidia Stock Want Some Of Your Money Off This Roller Coaster? Even Nvidia fell 66% at its worst in five years. Swings like this come with owning single stocks. The Trefis Income Solution holds 50 companies chosen to keep paying dividends, each with at least 10 straight years of dividends, yielding about 3x the S&P 500. See if it fits your plan. For questions about this article or underlying research, email [email protected]. Terms and Privacy Policy Your Privacy Choices CA Privacy Notice More Info
Anthropic IPO: Buy, Sell Or Hold Nvidia Stock?
More on the Anthropic IPO: our Anthropic IPO storyline page collects our pieces on it.
Tonight you might ask Claude to fix a broken formula in the family budget sheet. The answer arrives in seconds, worked out on a chip you will never see. Which chip matters to anyone who owns Nvidia. Anthropic's IPO filing says it runs Claude on four chip families: Amazon's Trainium, Google's TPUs, and GPUs from Nvidia and AMD.
Everyone knows this week's Nvidia story: a Wednesday close of $173.73, 27.4% below its 52-week closing high. That evening, The Wall Street Journal reported the board was preparing to oust CEO Jensen Huang; shares fell about 8% after hours, and no decision had been announced by early Thursday (see Nvidia Stock At $174: Sell After An 18% Drop?).
Fewer have read where Anthropic's compute money goes. Its filing ties about $254 billion of named deals to Nvidia GPUs, and $155.4 billion of it sits in deals that can be ended on short notice by early 2027. Its Google and Amazon deals cannot. Our answer: hold Nvidia, sized from its worst drop.
That $155 billion is the swing money. In a seven-figure portfolio, index funds hold Nvidia, Alphabet and Amazon side by side, so where it lands moves several holdings. Which way does yours tilt? A free 15-minute call with a Trefis partner firm can map it.
Where Does Anthropic's Compute Money Go?
| Chips | Named deals | Easy to exit? |
|---|---|---|
| Google TPUs | $282.5B (Google Cloud $121.3B, Broadcom leases $161.2B) | No |
| More TPUs, 2028-2031 | $125B-$150B, estimated | Not yet fixed |
| Amazon AWS | $110.0B or more | No |
| Nvidia GPUs | About $254B (SpaceXAI up to $120.5B, Microsoft $47.2B+, Nscale up to $44.6B, Lambda $41.3B) | $155.4B of it, yes |
| AMD GPUs | More than $20B | No |
Easy to exit?
$282.5B (Google Cloud $121.3B, Broadcom leases $161.2B)
More TPUs, 2028-2031
$125B-$150B, estimated
Not yet fixed
$110.0B or more
About $254B (SpaceXAI up to $120.5B, Microsoft $47.2B+, Nscale up to $44.6B, Lambda $41.3B)
$155.4B of it, yes
More than $20B
Source: Anthropic S-1/A, Nov 10; sums are ours. SpaceXAI deals become cancellable on 90 days' notice from early 2027; a $34.9B Lambda deal is terminable now (Anthropic doesn't plan to end it).
Google's TPUs and Amazon's AWS: $392.5 billion of fixed money. Our Nov 2 piece, Anthropic Has Promised $417 Billion For Computing, counted contracts signed January to May in the first filing; the amended one puts total commitments near $900.9 billion.
Is That A Problem For Nvidia's Sales?
| Nvidia vs Anthropic | Number |
|---|---|
| Nvidia data center sales , quarter to July 26 | $89B, +18% on the quarter |
| Anthropic contracted payments due in 12 months after Sept. 30 , all suppliers | $70.7B |
| Nvidia's Anthropic purchase at the IPO | $10.0B, about 0.75% of Anthropic, fully diluted |
Nvidia vs Anthropic
Nvidia data center sales, quarter to July 26
$89B, +18% on the quarter
Anthropic contracted payments due in 12 months after Sept. 30, all suppliers
Nvidia's Anthropic purchase at the IPO
$10.0B, about 0.75% of Anthropic, fully diluted
Anthropic's contracted 12-month payments, every supplier included, are less than Nvidia's data center sales in a single quarter. Sources: Nvidia's Aug 26 call; Anthropic S-1/A.
Asked in August about labs building their own chips, Huang said he expects them to keep using Nvidia for much of their computing.
Has Nvidia Paid For Its Pain?
| 5 years to Nov 11 | Return a year | Worst drop | Gain per pain |
|---|---|---|---|
| Nvidia | +41.9% | −66.3% | 1.41 |
| Broadcom | +46.7% | −41.1% | 2.32 |
| Alphabet | +15.2% | −44.3% | 0.61 |
| S&P 500 (SPY) | +11.9% | −24.5% | 0.87 |
5 years to Nov 11
Return a year
Gain per pain
S&P 500 (SPY)
Gain per pain: yearly return above T-bills (3.8%), divided by the typical size of down months. Above 1 is good. Source: Trefis, adjusted prices through Nov 11.
So Buy, Sell Or Hold Nvidia This Week?
Size it from the worst drop. With a 1-3% loss limit per stock, a repeat of the 66% fall means holding Nvidia at about 1.5-4.5% of everything you own, index funds included.
Watch dated triggers. Nov 17: Nvidia's revenue against its $108 billion guide. About Nov 18: Anthropic's IPO price. Early 2027: the SpaceXAI deals become cancellable. And any board announcement on Huang.
What it replaces: selling Nvidia on an IPO headline. The risk: other big labs copy the four-chip setup, and more flexible money moves to custom chips.
Why Would Nvidia Buy Into A Customer That Shops Around?
Anthropic's four-chip setup cuts its dependence on any one vendor. Nvidia's answer: own part of the customer. Its $10 billion is about 0.24% of Nvidia's $4.19 trillion market value, a small price for a seat at the table.
What Should You Ask Next?
- Anthropic IPO: What It Means For Amazon, Google And Nvidia Stock
Want Some Of Your Money Off This Roller Coaster?
Even Nvidia fell 66% at its worst in five years. Swings like this come with owning single stocks. The Trefis Income Solution holds 50 companies chosen to keep paying dividends, each with at least 10 straight years of dividends, yielding about 3x the S&P 500. See if it fits your plan.
For questions about this article or underlying research, email [email protected].
- Anthropic says a key AI safety alarm has been triggered, days before its IPO Jared Blikre Tue, November 10, 2026 at 10:00 PM PST 4 min read ANTH.PVT A week ago, Anthropic told would-be investors that AI speeding up its own research might "imminently" trigger one of the company's key safety alarms. On Tuesday, in the same filing that set a price range for its record-setting IPO, it said that has now happened. "We made a determination that we have met this threshold," the company wrote in its amended prospectus. It based the call on "what we expect from future models and our overall rate of progress, rather than on what we have observed from current models." The safety threshold sits in Anthropic's Responsible Scaling Policy — the company rulebook for adding tougher safety measures as its AI gets more powerful. This one is triggered when AI is doing enough of Anthropic's research to roughly double how fast its models improve. The policy says that pace would pack two years' worth of AI progress, at its 2018-2024 rate, into a single year. The filing says AI "has already been accelerating our research and development." Crossing the line means more safeguards, not a shutdown. The fear is AI will improve itself faster than people can check it. In 2023, Anthropic CEO Dario Amodei joined other AI leaders in warning that the "risk of extinction from AI" should be a global priority alongside pandemics and nuclear war. Anthropic says humans stay in charge. It doesn't expect to safely use Claude "to fully autonomously design and build successive generations of frontier models without meaningful human supervision or understanding," the filing says, and it is weighing policy changes "to keep humans in control." An outside group sees the same speedup. METR, an independent nonprofit that tests AI systems, estimated last week that Anthropic's models have been improving nearly twice as fast since October 2025. METR, Yahoo Finance In a separate review, METR called Anthropic's August argument that it was still below the threshold "overstated" and said it was "plausible" the line was crossed by mid-July. A $1.4 trillion debut Anthropic set its price range at $70 to $78 per share on Tuesday as it began its roadshow, the round of investor meetings before pricing. That values it at as much as about $1.4 trillion, counting every share that could exist. Trading could start as soon as Nov. 19, according to reports. Read more: Everything you need to know ahead of Anthropic's IPO Some of the market's biggest names are tied up in the IPO. Nvidia (NVDA) and Alphabet's (GOOG, GOOGL) Google are each buying $10 billion of Anthropic stock at the IPO price, part of a $26 billion private placement with AMD (AMD), Broadcom (AVGO), Micron (MU), Microsoft (MSFT), and Taiwan Semiconductor (TSM). Amazon (AMZN) and Google already own big stakes. For investors, the risk factors, where companies list everything that could go wrong, paint a darker picture. They warn that after a catastrophic AI incident, "no single insurer, and potentially the insurance industry in the aggregate," may have the capacity to cover the damage. Not everyone takes the warnings at face value. "As in all S-1 filings, the risks are deliberately overstated," Constellation Research founder Ray Wang said ahead of a Yahoo Finance live show appearance last month. A new name for AI The same filing carries a quieter change. Anthropic swapped in "super intelligence" for "AI" in places, citing President Trump's Sept. 29 executive order telling federal agencies to use the term. Mentions of "AI" fell by about a quarter from the Nov. 2 version, by Yahoo Finance's count, though the filing treats the terms as interchangeable. It came two weeks after Trump knocked the Wall Street Journal on Truth Social for sticking with the phrase "artificial intelligence." Anthropic SEC filings, Yahoo FinanceAs for the safety threshold, Anthropic said it is still weighing what crossing it means and will release further reports. It also flagged a bigger risk. In the filing's words, "As we increasingly leverage Claude to accelerate progress within Anthropic, our organization may lose visibility across certain processes in a way that may become difficult for any individual or group of humans to understand and control." Internal AI tools helped analyze interview transcripts, research, and draft this post. Humans reviewed, edited, and fact-checked this post before publication. Follow Jared Blikre on X at @SPYJared or email him at [email protected]. Terms and Privacy Policy Your Privacy Choices CA Privacy Notice More Info {"uid":"b","hostPeerName":"https://finance.yahoo.com","initialGeometry":"{\"windowCoords_t\":0,\"windowCoords_r\":1440,\"windowCoords_b\":1800,\"windowCoords_l\":0,\"frameCoords_t\":12690.40625,\"frameCoords_r\":862,\"frameCoords_b\":12940.40625,\"frameCoords_l\":222,\"styleZIndex\":\"auto\",\"allowedExpansion_t\":0,\"allowedExpansion_r\":0,\"allowedExpansion_b\":0,\"allowedExpansion_l\":0,\"xInView\":0,\"yInView\":0}","permissions":"{\"expandByOverlay\":false,\"expandByPush\":false,\"readCookie\":false,\"writeCookie\":false}","metadata":"{\"shared\":{\"sf_ver\":\"1-0-45\",\"ck_on\":1,\"flash_ver\":\"0\"}}","reportCreativeGeometry":false,"isDifferentSourceWindow":false,"goog_safeframe_hlt":{}}">
Anthropic says a key AI safety alarm has been triggered, days before its IPO
A week ago, Anthropic told would-be investors that AI speeding up its own research might "imminently" trigger one of the company's key safety alarms. On Tuesday, in the same filing that set a price range for its record-setting IPO, it said that has now happened.
"We made a determination that we have met this threshold," the company wrote in its amended prospectus. It based the call on "what we expect from future models and our overall rate of progress, rather than on what we have observed from current models."
The safety threshold sits in Anthropic's Responsible Scaling Policy — the company rulebook for adding tougher safety measures as its AI gets more powerful. This one is triggered when AI is doing enough of Anthropic's research to roughly double how fast its models improve. The policy says that pace would pack two years' worth of AI progress, at its 2018-2024 rate, into a single year. The filing says AI "has already been accelerating our research and development." Crossing the line means more safeguards, not a shutdown.
The fear is AI will improve itself faster than people can check it. In 2023, Anthropic CEO Dario Amodei joined other AI leaders in warning that the "risk of extinction from AI" should be a global priority alongside pandemics and nuclear war.
Anthropic says humans stay in charge. It doesn't expect to safely use Claude "to fully autonomously design and build successive generations of frontier models without meaningful human supervision or understanding," the filing says, and it is weighing policy changes "to keep humans in control."
An outside group sees the same speedup. METR, an independent nonprofit that tests AI systems, estimated last week that Anthropic's models have been improving nearly twice as fast since October 2025.
In a separate review, METR called Anthropic's August argument that it was still below the threshold "overstated" and said it was "plausible" the line was crossed by mid-July.
A $1.4 trillion debut
Anthropic set its price range at $70 to $78 per share on Tuesday as it began its roadshow, the round of investor meetings before pricing. That values it at as much as about $1.4 trillion, counting every share that could exist. Trading could start as soon as Nov. 19, according to reports.
Read more: Everything you need to know ahead of Anthropic's IPO
Some of the market's biggest names are tied up in the IPO. Nvidia (NVDA) and Alphabet's (GOOG, GOOGL) Google are each buying $10 billion of Anthropic stock at the IPO price, part of a $26 billion private placement with AMD (AMD), Broadcom (AVGO), Micron (MU), Microsoft (MSFT), and Taiwan Semiconductor (TSM). Amazon (AMZN) and Google already own big stakes.
For investors, the risk factors, where companies list everything that could go wrong, paint a darker picture. They warn that after a catastrophic AI incident, "no single insurer, and potentially the insurance industry in the aggregate," may have the capacity to cover the damage.
Not everyone takes the warnings at face value. "As in all S-1 filings, the risks are deliberately overstated," Constellation Research founder Ray Wang said ahead of a Yahoo Finance live show appearance last month.
A new name for AI
The same filing carries a quieter change. Anthropic swapped in "super intelligence" for "AI" in places, citing President Trump's Sept. 29 executive order telling federal agencies to use the term. Mentions of "AI" fell by about a quarter from the Nov. 2 version, by Yahoo Finance's count, though the filing treats the terms as interchangeable.
It came two weeks after Trump knocked the Wall Street Journal on Truth Social for sticking with the phrase "artificial intelligence."
As for the safety threshold, Anthropic said it is still weighing what crossing it means and will release further reports. It also flagged a bigger risk.
In the filing's words, "As we increasingly leverage Claude to accelerate progress within Anthropic, our organization may lose visibility across certain processes in a way that may become difficult for any individual or group of humans to understand and control."
Internal AI tools helped analyze interview transcripts, research, and draft this post. Humans reviewed, edited, and fact-checked this post before publication.
Follow Jared Blikre on X at @SPYJared or email him at [email protected].
- Claude can now join video calls as an active, visible teammate in Anthropic’s latest push to reinvent white collar jobs beyond coding Emily Forlini Tue, November 10, 2026 at 11:07 AM PST 4 min read ANTH.PVT Anthropic 's Claude Code reinvented the field of software engineering, turning AI into a must-have tool for programmers. Now the company is seeking to do the same across the corporate workplace with a new suite of office productivity tools launched on Tuesday. The new products are an important part of Anthropic's effort to broaden its base of paying customers and boost its revenue as it moves closer to a widely-anticipated initial public stock offering. The IPO, which could value Anthropic at roughly $1.4 trillion, is scheduled for as soon as next week, the company revealed separately on Tuesday. {"uid":"f","hostPeerName":"https://finance.yahoo.com","initialGeometry":"{\"windowCoords_t\":0,\"windowCoords_r\":1440,\"windowCoords_b\":1800,\"windowCoords_l\":0,\"frameCoords_t\":13568.390625,\"frameCoords_r\":862,\"frameCoords_b\":13818.390625,\"frameCoords_l\":222,\"styleZIndex\":\"auto\",\"allowedExpansion_t\":0,\"allowedExpansion_r\":0,\"allowedExpansion_b\":0,\"allowedExpansion_l\":0,\"xInView\":0,\"yInView\":0}","permissions":"{\"expandByOverlay\":false,\"expandByPush\":false,\"readCookie\":false,\"writeCookie\":false}","metadata":"{\"shared\":{\"sf_ver\":\"1-0-45\",\"ck_on\":1,\"flash_ver\":\"0\"}}","reportCreativeGeometry":false,"isDifferentSourceWindow":false,"goog_safeframe_hlt":{}}" id="google_ads_iframe_/22888152279/us/yfin/ros/dt/us_yfin_ros_dt_mid_center_4" scrolling="no" title="3rd party ad content" height="250" src="https://01f62a58257ce51d82d039787788c17a.safeframe.googlesyndication.com/safeframe/1-0-45/html/container.html" marginwidth="0" allow="private-state-token-redemption" marginheight="0"> To help create momentum around the product launch, Anthropic held a buzzy launch event in San Francisco Tuesday with hundreds of guests. Anthropic's Chief Product Officer Ami Vora spoke on stage about how Claude has changed computer coding over the past year. Engineers can now delegate entire tasks to AI, but at the same time the work for most other teams has largely stayed the same, save for perhaps an email getting written faster, or a meeting being summarized. The new products aim to bring the same AI benefits programmers have enjoyed "for over a year," Vora said, to knowledge workers across corporations. Anthropic's renewed push to attract corporate knowledge workers could put the company in closer competition with other software vendors such as Microsoft, whose shares were down more than 2% after the news was announced. Anthropic's vision is to turn Claude into an active member a team, rather than an optimization tool for individuals. For example, with Claude Meetings, professionals can add Claude to video calls as a "visible participant," Anthropic said. The AI listens to the conversation and can "start on the work while people are still talking." It lays out the proposed plan and timeline, answers questions in the call chat, and handles tasks such as pulling up last year's numbers, using "the work apps people have connected," Anthropic said. The results surface in claude.ai after the call. Claude Meetings is in beta for Team and Enterprise plans, and starting with Google Meet, with plans to expand to other video platforms. With Claude Whiteboards, humans can brainstorm ideas and collaborate with Claude. "A team talks through a messy plan out loud, and Claude turns it into a board with a timeline, owners and dates," Anthropic said. Same goes for Claude Sheets, which are spreadsheets teams can build and edit alongside Claude, and can export to Excel or CSV files. Both these products are also in beta. If these products can automate a meaningful portion of non-coding roles, they could encourage more businesses to sign up for Claude subscription plans. However, they could also create mayhem or burn tokens on AI-generated materials that never get used. Anthropic is also seeking ways to deeply embed Claude in critical business functions. Another product called Claude Data, formerly Claude Dashboards, connects the AI to systems where companies keep their most important numbers, such as a data warehouse or CRM. Claude Data became available for paid plans today. Similarly, new Claude Connectors "link Claude to the apps a company already uses, so Claude can pull information from them and take actions in them," Anthropic said. Over 9,000 apps are available as Claude Connectors, such as Salesforce and Workday. "Teams already do much of their work in apps like these," Anthropic said. "Now they can open the full app beside the conversation and ask Claude to work in it with them." That means "no more switching tabs," said Nicholas Lin, who leads product for knowledge work at Anthropic, on stage at today's event. The road to the IPO Tuesday's launch event came on the same day as Anthropic kicked off its pre-IPO roadshow with investors. The idea of Claude revolutionizing non-coding roles in the same way it's done for coding could energize the IPO and drive up the share price. Several sources we spoke to think Anthropic's $70-$78 target share price is intentionally low, a tactic to set expectations and leave room to shatter them. "Anthropic is moving forward with its IPO at a steeper discount than market initially anticipated," said Lior Prosor, managing partner of Deep33, a deep-tech venture fund. "We think that is a smart pricing strategy. It gives nervous 'AI-trade' investors meaningful potential upside while securing the capital Anthropic needs to take full control of its destiny and fund its compute commitments." Also this week, Anthropic debuted new frontier models, Fable 5.2 and Mythos 5.2, at a lower price with allegedly stronger performance than OpenAI's latest frontier model, GPT-6 Astra. This also got investors' attention. "Anthropic has shown incredible advances recently and their announcement about Fable 5.2 yesterday suggests they could have a bigger impact in hardware domains like robotics than people might have given them credit for previously," said Bastiaan Janmaat, General Partner at Lithos Capital. This story was originally featured on Fortune.com
Claude can now join video calls as an active, visible teammate in Anthropic’s latest push to reinvent white collar jobs beyond coding
Anthropic 's Claude Code reinvented the field of software engineering, turning AI into a must-have tool for programmers. Now the company is seeking to do the same across the corporate workplace with a new suite of office productivity tools launched on Tuesday.
The new products are an important part of Anthropic's effort to broaden its base of paying customers and boost its revenue as it moves closer to a widely-anticipated initial public stock offering. The IPO, which could value Anthropic at roughly $1.4 trillion, is scheduled for as soon as next week, the company revealed separately on Tuesday.
To help create momentum around the product launch, Anthropic held a buzzy launch event in San Francisco Tuesday with hundreds of guests. Anthropic's Chief Product Officer Ami Vora spoke on stage about how Claude has changed computer coding over the past year. Engineers can now delegate entire tasks to AI, but at the same time the work for most other teams has largely stayed the same, save for perhaps an email getting written faster, or a meeting being summarized.
The new products aim to bring the same AI benefits programmers have enjoyed "for over a year," Vora said, to knowledge workers across corporations. Anthropic's renewed push to attract corporate knowledge workers could put the company in closer competition with other software vendors such as Microsoft, whose shares were down more than 2% after the news was announced.
Anthropic's vision is to turn Claude into an active member a team, rather than an optimization tool for individuals. For example, with Claude Meetings, professionals can add Claude to video calls as a "visible participant," Anthropic said. The AI listens to the conversation and can "start on the work while people are still talking." It lays out the proposed plan and timeline, answers questions in the call chat, and handles tasks such as pulling up last year's numbers, using "the work apps people have connected," Anthropic said. The results surface in claude.ai after the call.
Claude Meetings is in beta for Team and Enterprise plans, and starting with Google Meet, with plans to expand to other video platforms.
With Claude Whiteboards, humans can brainstorm ideas and collaborate with Claude. "A team talks through a messy plan out loud, and Claude turns it into a board with a timeline, owners and dates," Anthropic said. Same goes for Claude Sheets, which are spreadsheets teams can build and edit alongside Claude, and can export to Excel or CSV files. Both these products are also in beta.
If these products can automate a meaningful portion of non-coding roles, they could encourage more businesses to sign up for Claude subscription plans. However, they could also create mayhem or burn tokens on AI-generated materials that never get used.
Anthropic is also seeking ways to deeply embed Claude in critical business functions. Another product called Claude Data, formerly Claude Dashboards, connects the AI to systems where companies keep their most important numbers, such as a data warehouse or CRM. Claude Data became available for paid plans today.
Similarly, new Claude Connectors "link Claude to the apps a company already uses, so Claude can pull information from them and take actions in them," Anthropic said. Over 9,000 apps are available as Claude Connectors, such as Salesforce and Workday. "Teams already do much of their work in apps like these," Anthropic said. "Now they can open the full app beside the conversation and ask Claude to work in it with them." That means "no more switching tabs," said Nicholas Lin, who leads product for knowledge work at Anthropic, on stage at today's event.
The road to the IPO
Tuesday's launch event came on the same day as Anthropic kicked off its pre-IPO roadshow with investors.
The idea of Claude revolutionizing non-coding roles in the same way it's done for coding could energize the IPO and drive up the share price. Several sources we spoke to think Anthropic's $70-$78 target share price is intentionally low, a tactic to set expectations and leave room to shatter them.
"Anthropic is moving forward with its IPO at a steeper discount than market initially anticipated," said Lior Prosor, managing partner of Deep33, a deep-tech venture fund. "We think that is a smart pricing strategy. It gives nervous 'AI-trade' investors meaningful potential upside while securing the capital Anthropic needs to take full control of its destiny and fund its compute commitments."
Also this week, Anthropic debuted new frontier models, Fable 5.2 and Mythos 5.2, at a lower price with allegedly stronger performance than OpenAI's latest frontier model, GPT-6 Astra. This also got investors' attention.
"Anthropic has shown incredible advances recently and their announcement about Fable 5.2 yesterday suggests they could have a bigger impact in hardware domains like robotics than people might have given them credit for previously," said Bastiaan Janmaat, General Partner at Lithos Capital.
This story was originally featured on Fortune.com