News & Updates
The latest news and updates from companies in the WLTH portfolio.
Anthropic makes quiet move Nvidia investors must consider
Amir Salek spent four years doing something that looked a lot like walking away from the chip business. After founding and running Google's Tensor Processing Unit program from 2013 to 2022 and shipping seven generations of the custom silicon that now powers much of Google's AI infrastructure, he left to invest in deep-tech startups at Cerberus Capital Management, a press release announced at the time. That detour is why his next move matters more than a routine hire. Anthropic has brought Salek onto its compute team, where he will report to compute lead James Bradbury, according to Bloomberg. He is not being poached out of a rival's office. He is being pulled back into chip design after four years on the sidelines by a company that did not have a hardware team a year ago. Salek is not Anthropic's first hardware hire this year Two months earlier, Anthropic landed Clive Chan, the second engineer ever hired to OpenAI's custom chip program, who had spent more than two years building the Broadcom-designed inference accelerator OpenAI markets as Jalapeno. Chan announced the move himself in June, framing it as a step up rather than a rescue. The pattern is what makes the Salek hire read as strategy instead of opportunism. Anthropic publicly confirmed on Aug. 5 that it is standing up an in-house silicon team, with the explicit goal of co-designing chips and Claude models together to cut inference costs by roughly half, according to TechCrunch. TechCrunch also reported that Anthropic has held early manufacturing talks with Samsung. Two hires in three months, both pulled from the two companies furthest along in custom AI silicon, is not a coincidence. It is a shopping list. Javier Zayas Photography / Getty Images Raiding Google's bench Here is the part most coverage of the Salek hire has skipped past. Google is not a bystander in Anthropic's chip ambitions. It is the supplier. Anthropic agreed in October to buy up to a million of Google's Tensor Processing Units in a deal worth tens of billions of dollars, then expanded that arrangement in April to cover multiple additional gigawatts of TPU capacity through 2027, according to the company's announcement. Google is, in other words, simultaneously arming Anthropic with chips after losing the person most responsible for the chip program that those chips came from. That is an unusual position for a supplier to hold with a customer that competes directly against its own Gemini AI models. It complicates the idea that the relationship is simply commercial. Anthropic still says AWS Trainium, Google TPUs and Nvidia GPUs will remain central to how it scales Claude, and nothing about Salek's hire changes that in the short run. For investors in Nvidia, Alphabet and Broadcom, the practical impact today is limited, since custom silicon takes years to design and manufacture at volume. But each AI lab that builds its own silicon program shrinks the list of frontier customers still fully dependent on Nvidia. Broadcom, which already designs custom chips for OpenAI and Google, is the most likely beneficiary if Anthropic's effort advances that far. The real contest has moved from GPUs to headcount Bidding wars over AI researchers are old news by now. What's newer is the fight over hardware architects, the engineers who can take a chip from blueprint to production line. There are far fewer of them than machine learning researchers, and that scarcity is starting to show. That scarcity has already produced legal fights in other industries. Warner Bros. Discovery sued Amazon in July over an executive who left a fixed-term contract 16 months early, arguing Amazon ran what its complaint called a lawless hiring spree. Salek's move to anthropic carries none of that risk. He had already left Google's payroll years before Anthropic called, which means there is no contract to breach and no lawsuit to file. That is the quieter lesson here. The clean way to build a chip team is not to raid a competitor mid-contract. It is to identify who already built one of these programs once, wait until they are between roles, and make the offer before anyone else does. The custom silicon race Meta's in-house Iris chip is due in production by September. OpenAi's Jalapeno targets the back half of this year. Microsoft has run Maia chips for two years, and Amazon has done the same with Trainium. Anthropic is the last major lab into custom silicon, not the first. Owning hardware talent is not the same as owning working silicon. Anthropic's chip, whatever it becomes, is not expected before 2028 at the earliest. The next test is not who Anthropic hires next. It is whether a lab built on renting other companies' chips can actually ship one of its own. As frontier AI labs and hyperscalers double down on in-house silicon, a critical question emerges for Wall Street: Does this mark the beginning of structural market share erosion for Nvidia? While Nvidia's total addressable market continues to expand rapidly, every successful custom ASIC deployed for inference or specialized training shifts workloads away from general purpose GPUs, slowly chipping away at the revenue concentration of the market's leading chipmaker. The Arena Media Brands, LLC THESTREET is a registered trademark of TheStreet, Inc. This story was originally published August 23, 2026 at 4:33 PM.
Anthropic makes quiet move Nvidia investors must consider
Amir Salek spent four years doing something that looked a lot like walking away from the chip business. After founding and running Google's Tensor Processing Unit program from 2013 to 2022 and shipping seven generations of the custom silicon that now powers much of Google's AI infrastructure, he left to invest in deep-tech startups at Cerberus Capital Management, a press release announced at the time. That detour is why his next move matters more than a routine hire. Anthropic has brought Salek onto its compute team, where he will report to compute lead James Bradbury, according to Bloomberg. He is not being poached out of a rival's office. He is being pulled back into chip design after four years on the sidelines by a company that did not have a hardware team a year ago. Salek is not Anthropic's first hardware hire this year Two months earlier, Anthropic landed Clive Chan, the second engineer ever hired to OpenAI's custom chip program, who had spent more than two years building the Broadcom-designed inference accelerator OpenAI markets as Jalapeno. Chan announced the move himself in June, framing it as a step up rather than a rescue. The pattern is what makes the Salek hire read as strategy instead of opportunism. Anthropic publicly confirmed on Aug. 5 that it is standing up an in-house silicon team, with the explicit goal of co-designing chips and Claude models together to cut inference costs by roughly half, according to TechCrunch. TechCrunch also reported that Anthropic has held early manufacturing talks with Samsung. Two hires in three months, both pulled from the two companies furthest along in custom AI silicon, is not a coincidence. It is a shopping list. Javier Zayas Photography / Getty Images Raiding Google's bench Here is the part most coverage of the Salek hire has skipped past. Google is not a bystander in Anthropic's chip ambitions. It is the supplier. Anthropic agreed in October to buy up to a million of Google's Tensor Processing Units in a deal worth tens of billions of dollars, then expanded that arrangement in April to cover multiple additional gigawatts of TPU capacity through 2027, according to the company's announcement. Google is, in other words, simultaneously arming Anthropic with chips after losing the person most responsible for the chip program that those chips came from. That is an unusual position for a supplier to hold with a customer that competes directly against its own Gemini AI models. It complicates the idea that the relationship is simply commercial. Anthropic still says AWS Trainium, Google TPUs and Nvidia GPUs will remain central to how it scales Claude, and nothing about Salek's hire changes that in the short run. For investors in Nvidia, Alphabet and Broadcom, the practical impact today is limited, since custom silicon takes years to design and manufacture at volume. But each AI lab that builds its own silicon program shrinks the list of frontier customers still fully dependent on Nvidia. Broadcom, which already designs custom chips for OpenAI and Google, is the most likely beneficiary if Anthropic's effort advances that far. The real contest has moved from GPUs to headcount Bidding wars over AI researchers are old news by now. What's newer is the fight over hardware architects, the engineers who can take a chip from blueprint to production line. There are far fewer of them than machine learning researchers, and that scarcity is starting to show. That scarcity has already produced legal fights in other industries. Warner Bros. Discovery sued Amazon in July over an executive who left a fixed-term contract 16 months early, arguing Amazon ran what its complaint called a lawless hiring spree. Salek's move to anthropic carries none of that risk. He had already left Google's payroll years before Anthropic called, which means there is no contract to breach and no lawsuit to file. That is the quieter lesson here. The clean way to build a chip team is not to raid a competitor mid-contract. It is to identify who already built one of these programs once, wait until they are between roles, and make the offer before anyone else does. The custom silicon race Meta's in-house Iris chip is due in production by September. OpenAi's Jalapeno targets the back half of this year. Microsoft has run Maia chips for two years, and Amazon has done the same with Trainium. Anthropic is the last major lab into custom silicon, not the first. Owning hardware talent is not the same as owning working silicon. Anthropic's chip, whatever it becomes, is not expected before 2028 at the earliest. The next test is not who Anthropic hires next. It is whether a lab built on renting other companies' chips can actually ship one of its own. As frontier AI labs and hyperscalers double down on in-house silicon, a critical question emerges for Wall Street: Does this mark the beginning of structural market share erosion for Nvidia? While Nvidia's total addressable market continues to expand rapidly, every successful custom ASIC deployed for inference or specialized training shifts workloads away from general purpose GPUs, slowly chipping away at the revenue concentration of the market's leading chipmaker. The Arena Media Brands, LLC THESTREET is a registered trademark of TheStreet, Inc. This story was originally published August 23, 2026 at 5:33 PM.
Anthropic makes quiet move Nvidia investors must consider
Amir Salek spent four years doing something that looked a lot like walking away from the chip business. After founding and running Google's Tensor Processing Unit program from 2013 to 2022 and shipping seven generations of the custom silicon that now powers much of Google's AI infrastructure, he left to invest in deep-tech startups at Cerberus Capital Management, a press release announced at the time. That detour is why his next move matters more than a routine hire. Anthropic has brought Salek onto its compute team, where he will report to compute lead James Bradbury, according to Bloomberg. He is not being poached out of a rival's office. He is being pulled back into chip design after four years on the sidelines by a company that did not have a hardware team a year ago. Salek is not Anthropic's first hardware hire this year Two months earlier, Anthropic landed Clive Chan, the second engineer ever hired to OpenAI's custom chip program, who had spent more than two years building the Broadcom-designed inference accelerator OpenAI markets as Jalapeno. Chan announced the move himself in June, framing it as a step up rather than a rescue. The pattern is what makes the Salek hire read as strategy instead of opportunism. Anthropic publicly confirmed on Aug. 5 that it is standing up an in-house silicon team, with the explicit goal of co-designing chips and Claude models together to cut inference costs by roughly half, according to TechCrunch. TechCrunch also reported that Anthropic has held early manufacturing talks with Samsung. Two hires in three months, both pulled from the two companies furthest along in custom AI silicon, is not a coincidence. It is a shopping list. Javier Zayas Photography / Getty Images Raiding Google's bench Here is the part most coverage of the Salek hire has skipped past. Google is not a bystander in Anthropic's chip ambitions. It is the supplier. Anthropic agreed in October to buy up to a million of Google's Tensor Processing Units in a deal worth tens of billions of dollars, then expanded that arrangement in April to cover multiple additional gigawatts of TPU capacity through 2027, according to the company's announcement. Google is, in other words, simultaneously arming Anthropic with chips after losing the person most responsible for the chip program that those chips came from. That is an unusual position for a supplier to hold with a customer that competes directly against its own Gemini AI models. It complicates the idea that the relationship is simply commercial. Anthropic still says AWS Trainium, Google TPUs and Nvidia GPUs will remain central to how it scales Claude, and nothing about Salek's hire changes that in the short run. For investors in Nvidia, Alphabet and Broadcom, the practical impact today is limited, since custom silicon takes years to design and manufacture at volume. But each AI lab that builds its own silicon program shrinks the list of frontier customers still fully dependent on Nvidia. Broadcom, which already designs custom chips for OpenAI and Google, is the most likely beneficiary if Anthropic's effort advances that far. The real contest has moved from GPUs to headcount Bidding wars over AI researchers are old news by now. What's newer is the fight over hardware architects, the engineers who can take a chip from blueprint to production line. There are far fewer of them than machine learning researchers, and that scarcity is starting to show. That scarcity has already produced legal fights in other industries. Warner Bros. Discovery sued Amazon in July over an executive who left a fixed-term contract 16 months early, arguing Amazon ran what its complaint called a lawless hiring spree. Salek's move to anthropic carries none of that risk. He had already left Google's payroll years before Anthropic called, which means there is no contract to breach and no lawsuit to file. That is the quieter lesson here. The clean way to build a chip team is not to raid a competitor mid-contract. It is to identify who already built one of these programs once, wait until they are between roles, and make the offer before anyone else does. The custom silicon race Meta's in-house Iris chip is due in production by September. OpenAi's Jalapeno targets the back half of this year. Microsoft has run Maia chips for two years, and Amazon has done the same with Trainium. Anthropic is the last major lab into custom silicon, not the first. Owning hardware talent is not the same as owning working silicon. Anthropic's chip, whatever it becomes, is not expected before 2028 at the earliest. The next test is not who Anthropic hires next. It is whether a lab built on renting other companies' chips can actually ship one of its own. As frontier AI labs and hyperscalers double down on in-house silicon, a critical question emerges for Wall Street: Does this mark the beginning of structural market share erosion for Nvidia? While Nvidia's total addressable market continues to expand rapidly, every successful custom ASIC deployed for inference or specialized training shifts workloads away from general purpose GPUs, slowly chipping away at the revenue concentration of the market's leading chipmaker. The Arena Media Brands, LLC THESTREET is a registered trademark of TheStreet, Inc. This story was originally published August 23, 2026 at 4:33 PM.
Anthropic makes quiet move Nvidia investors must consider
Amir Salek spent four years doing something that looked a lot like walking away from the chip business. After founding and running Google's Tensor Processing Unit program from 2013 to 2022 and shipping seven generations of the custom silicon that now powers much of Google's AI infrastructure, he left to invest in deep-tech startups at Cerberus Capital Management, a press release announced at the time. That detour is why his next move matters more than a routine hire. Anthropic has brought Salek onto its compute team, where he will report to compute lead James Bradbury, according to Bloomberg. He is not being poached out of a rival's office. He is being pulled back into chip design after four years on the sidelines by a company that did not have a hardware team a year ago. Salek is not Anthropic's first hardware hire this year Two months earlier, Anthropic landed Clive Chan, the second engineer ever hired to OpenAI's custom chip program, who had spent more than two years building the Broadcom-designed inference accelerator OpenAI markets as Jalapeno. Chan announced the move himself in June, framing it as a step up rather than a rescue. The pattern is what makes the Salek hire read as strategy instead of opportunism. Anthropic publicly confirmed on Aug. 5 that it is standing up an in-house silicon team, with the explicit goal of co-designing chips and Claude models together to cut inference costs by roughly half, according to TechCrunch. TechCrunch also reported that Anthropic has held early manufacturing talks with Samsung. Two hires in three months, both pulled from the two companies furthest along in custom AI silicon, is not a coincidence. It is a shopping list. Javier Zayas Photography / Getty Images Raiding Google's bench Here is the part most coverage of the Salek hire has skipped past. Google is not a bystander in Anthropic's chip ambitions. It is the supplier. Anthropic agreed in October to buy up to a million of Google's Tensor Processing Units in a deal worth tens of billions of dollars, then expanded that arrangement in April to cover multiple additional gigawatts of TPU capacity through 2027, according to the company's announcement. Google is, in other words, simultaneously arming Anthropic with chips after losing the person most responsible for the chip program that those chips came from. That is an unusual position for a supplier to hold with a customer that competes directly against its own Gemini AI models. It complicates the idea that the relationship is simply commercial. Anthropic still says AWS Trainium, Google TPUs and Nvidia GPUs will remain central to how it scales Claude, and nothing about Salek's hire changes that in the short run. For investors in Nvidia, Alphabet and Broadcom, the practical impact today is limited, since custom silicon takes years to design and manufacture at volume. But each AI lab that builds its own silicon program shrinks the list of frontier customers still fully dependent on Nvidia. Broadcom, which already designs custom chips for OpenAI and Google, is the most likely beneficiary if Anthropic's effort advances that far. The real contest has moved from GPUs to headcount Bidding wars over AI researchers are old news by now. What's newer is the fight over hardware architects, the engineers who can take a chip from blueprint to production line. There are far fewer of them than machine learning researchers, and that scarcity is starting to show. That scarcity has already produced legal fights in other industries. Warner Bros. Discovery sued Amazon in July over an executive who left a fixed-term contract 16 months early, arguing Amazon ran what its complaint called a lawless hiring spree. Salek's move to anthropic carries none of that risk. He had already left Google's payroll years before Anthropic called, which means there is no contract to breach and no lawsuit to file. That is the quieter lesson here. The clean way to build a chip team is not to raid a competitor mid-contract. It is to identify who already built one of these programs once, wait until they are between roles, and make the offer before anyone else does. The custom silicon race Meta's in-house Iris chip is due in production by September. OpenAi's Jalapeno targets the back half of this year. Microsoft has run Maia chips for two years, and Amazon has done the same with Trainium. Anthropic is the last major lab into custom silicon, not the first. Owning hardware talent is not the same as owning working silicon. Anthropic's chip, whatever it becomes, is not expected before 2028 at the earliest. The next test is not who Anthropic hires next. It is whether a lab built on renting other companies' chips can actually ship one of its own. As frontier AI labs and hyperscalers double down on in-house silicon, a critical question emerges for Wall Street: Does this mark the beginning of structural market share erosion for Nvidia? While Nvidia's total addressable market continues to expand rapidly, every successful custom ASIC deployed for inference or specialized training shifts workloads away from general purpose GPUs, slowly chipping away at the revenue concentration of the market's leading chipmaker. The Arena Media Brands, LLC THESTREET is a registered trademark of TheStreet, Inc. This story was originally published August 23, 2026 at 7:33 PM.
Anthropic makes quiet move Nvidia investors must consider
Amir Salek spent four years doing something that looked a lot like walking away from the chip business. After founding and running Google's Tensor Processing Unit program from 2013 to 2022 and shipping seven generations of the custom silicon that now powers much of Google's AI infrastructure, he left to invest in deep-tech startups at Cerberus Capital Management, a press release announced at the time. That detour is why his next move matters more than a routine hire. Anthropic has brought Salek onto its compute team, where he will report to compute lead James Bradbury, according to Bloomberg. He is not being poached out of a rival's office. He is being pulled back into chip design after four years on the sidelines by a company that did not have a hardware team a year ago. Salek is not Anthropic's first hardware hire this year Two months earlier, Anthropic landed Clive Chan, the second engineer ever hired to OpenAI's custom chip program, who had spent more than two years building the Broadcom-designed inference accelerator OpenAI markets as Jalapeno. Chan announced the move himself in June, framing it as a step up rather than a rescue. The pattern is what makes the Salek hire read as strategy instead of opportunism. Anthropic publicly confirmed on Aug. 5 that it is standing up an in-house silicon team, with the explicit goal of co-designing chips and Claude models together to cut inference costs by roughly half, according to TechCrunch. TechCrunch also reported that Anthropic has held early manufacturing talks with Samsung. Two hires in three months, both pulled from the two companies furthest along in custom AI silicon, is not a coincidence. It is a shopping list. Javier Zayas Photography / Getty Images Raiding Google's bench Here is the part most coverage of the Salek hire has skipped past. Google is not a bystander in Anthropic's chip ambitions. It is the supplier. Anthropic agreed in October to buy up to a million of Google's Tensor Processing Units in a deal worth tens of billions of dollars, then expanded that arrangement in April to cover multiple additional gigawatts of TPU capacity through 2027, according to the company's announcement. Google is, in other words, simultaneously arming Anthropic with chips after losing the person most responsible for the chip program that those chips came from. That is an unusual position for a supplier to hold with a customer that competes directly against its own Gemini AI models. It complicates the idea that the relationship is simply commercial. Anthropic still says AWS Trainium, Google TPUs and Nvidia GPUs will remain central to how it scales Claude, and nothing about Salek's hire changes that in the short run. For investors in Nvidia, Alphabet and Broadcom, the practical impact today is limited, since custom silicon takes years to design and manufacture at volume. But each AI lab that builds its own silicon program shrinks the list of frontier customers still fully dependent on Nvidia. Broadcom, which already designs custom chips for OpenAI and Google, is the most likely beneficiary if Anthropic's effort advances that far. The real contest has moved from GPUs to headcount Bidding wars over AI researchers are old news by now. What's newer is the fight over hardware architects, the engineers who can take a chip from blueprint to production line. There are far fewer of them than machine learning researchers, and that scarcity is starting to show. That scarcity has already produced legal fights in other industries. Warner Bros. Discovery sued Amazon in July over an executive who left a fixed-term contract 16 months early, arguing Amazon ran what its complaint called a lawless hiring spree. Salek's move to anthropic carries none of that risk. He had already left Google's payroll years before Anthropic called, which means there is no contract to breach and no lawsuit to file. That is the quieter lesson here. The clean way to build a chip team is not to raid a competitor mid-contract. It is to identify who already built one of these programs once, wait until they are between roles, and make the offer before anyone else does. The custom silicon race Meta's in-house Iris chip is due in production by September. OpenAi's Jalapeno targets the back half of this year. Microsoft has run Maia chips for two years, and Amazon has done the same with Trainium. Anthropic is the last major lab into custom silicon, not the first. Owning hardware talent is not the same as owning working silicon. Anthropic's chip, whatever it becomes, is not expected before 2028 at the earliest. The next test is not who Anthropic hires next. It is whether a lab built on renting other companies' chips can actually ship one of its own. As frontier AI labs and hyperscalers double down on in-house silicon, a critical question emerges for Wall Street: Does this mark the beginning of structural market share erosion for Nvidia? While Nvidia's total addressable market continues to expand rapidly, every successful custom ASIC deployed for inference or specialized training shifts workloads away from general purpose GPUs, slowly chipping away at the revenue concentration of the market's leading chipmaker. The Arena Media Brands, LLC THESTREET is a registered trademark of TheStreet, Inc. This story was originally published August 23, 2026 at 7:33 PM.
Anthropic makes quiet move Nvidia investors must consider
Amir Salek spent four years doing something that looked a lot like walking away from the chip business. After founding and running Google's Tensor Processing Unit program from 2013 to 2022 and shipping seven generations of the custom silicon that now powers much of Google's AI infrastructure, he left to invest in deep-tech startups at Cerberus Capital Management, a press release announced at the time. That detour is why his next move matters more than a routine hire. Anthropic has brought Salek onto its compute team, where he will report to compute lead James Bradbury, according to Bloomberg. He is not being poached out of a rival's office. He is being pulled back into chip design after four years on the sidelines by a company that did not have a hardware team a year ago. Salek is not Anthropic's first hardware hire this year Two months earlier, Anthropic landed Clive Chan, the second engineer ever hired to OpenAI's custom chip program, who had spent more than two years building the Broadcom-designed inference accelerator OpenAI markets as Jalapeno. Chan announced the move himself in June, framing it as a step up rather than a rescue. The pattern is what makes the Salek hire read as strategy instead of opportunism. Anthropic publicly confirmed on Aug. 5 that it is standing up an in-house silicon team, with the explicit goal of co-designing chips and Claude models together to cut inference costs by roughly half, according to TechCrunch. TechCrunch also reported that Anthropic has held early manufacturing talks with Samsung. Two hires in three months, both pulled from the two companies furthest along in custom AI silicon, is not a coincidence. It is a shopping list. Javier Zayas Photography / Getty Images Raiding Google's bench Here is the part most coverage of the Salek hire has skipped past. Google is not a bystander in Anthropic's chip ambitions. It is the supplier. Anthropic agreed in October to buy up to a million of Google's Tensor Processing Units in a deal worth tens of billions of dollars, then expanded that arrangement in April to cover multiple additional gigawatts of TPU capacity through 2027, according to the company's announcement. Google is, in other words, simultaneously arming Anthropic with chips after losing the person most responsible for the chip program that those chips came from. That is an unusual position for a supplier to hold with a customer that competes directly against its own Gemini AI models. It complicates the idea that the relationship is simply commercial. Anthropic still says AWS Trainium, Google TPUs and Nvidia GPUs will remain central to how it scales Claude, and nothing about Salek's hire changes that in the short run. For investors in Nvidia, Alphabet and Broadcom, the practical impact today is limited, since custom silicon takes years to design and manufacture at volume. But each AI lab that builds its own silicon program shrinks the list of frontier customers still fully dependent on Nvidia. Broadcom, which already designs custom chips for OpenAI and Google, is the most likely beneficiary if Anthropic's effort advances that far. The real contest has moved from GPUs to headcount Bidding wars over AI researchers are old news by now. What's newer is the fight over hardware architects, the engineers who can take a chip from blueprint to production line. There are far fewer of them than machine learning researchers, and that scarcity is starting to show. That scarcity has already produced legal fights in other industries. Warner Bros. Discovery sued Amazon in July over an executive who left a fixed-term contract 16 months early, arguing Amazon ran what its complaint called a lawless hiring spree. Salek's move to anthropic carries none of that risk. He had already left Google's payroll years before Anthropic called, which means there is no contract to breach and no lawsuit to file. That is the quieter lesson here. The clean way to build a chip team is not to raid a competitor mid-contract. It is to identify who already built one of these programs once, wait until they are between roles, and make the offer before anyone else does. The custom silicon race Meta's in-house Iris chip is due in production by September. OpenAi's Jalapeno targets the back half of this year. Microsoft has run Maia chips for two years, and Amazon has done the same with Trainium. Anthropic is the last major lab into custom silicon, not the first. Owning hardware talent is not the same as owning working silicon. Anthropic's chip, whatever it becomes, is not expected before 2028 at the earliest. The next test is not who Anthropic hires next. It is whether a lab built on renting other companies' chips can actually ship one of its own. As frontier AI labs and hyperscalers double down on in-house silicon, a critical question emerges for Wall Street: Does this mark the beginning of structural market share erosion for Nvidia? While Nvidia's total addressable market continues to expand rapidly, every successful custom ASIC deployed for inference or specialized training shifts workloads away from general purpose GPUs, slowly chipping away at the revenue concentration of the market's leading chipmaker. The Arena Media Brands, LLC THESTREET is a registered trademark of TheStreet, Inc. This story was originally published August 23, 2026 at 4:33 PM.
Anthropic makes quiet move Nvidia investors must consider
Amir Salek spent four years doing something that looked a lot like walking away from the chip business. After founding and running Google's Tensor Processing Unit program from 2013 to 2022 and shipping seven generations of the custom silicon that now powers much of Google's AI infrastructure, he left to invest in deep-tech startups at Cerberus Capital Management, a press release announced at the time. That detour is why his next move matters more than a routine hire. Anthropic has brought Salek onto its compute team, where he will report to compute lead James Bradbury, according to Bloomberg. He is not being poached out of a rival's office. He is being pulled back into chip design after four years on the sidelines by a company that did not have a hardware team a year ago. Salek is not Anthropic's first hardware hire this year Two months earlier, Anthropic landed Clive Chan, the second engineer ever hired to OpenAI's custom chip program, who had spent more than two years building the Broadcom-designed inference accelerator OpenAI markets as Jalapeno. Chan announced the move himself in June, framing it as a step up rather than a rescue. The pattern is what makes the Salek hire read as strategy instead of opportunism. Anthropic publicly confirmed on Aug. 5 that it is standing up an in-house silicon team, with the explicit goal of co-designing chips and Claude models together to cut inference costs by roughly half, according to TechCrunch. TechCrunch also reported that Anthropic has held early manufacturing talks with Samsung. Two hires in three months, both pulled from the two companies furthest along in custom AI silicon, is not a coincidence. It is a shopping list. Javier Zayas Photography / Getty Images Raiding Google's bench Here is the part most coverage of the Salek hire has skipped past. Google is not a bystander in Anthropic's chip ambitions. It is the supplier. Anthropic agreed in October to buy up to a million of Google's Tensor Processing Units in a deal worth tens of billions of dollars, then expanded that arrangement in April to cover multiple additional gigawatts of TPU capacity through 2027, according to the company's announcement. Google is, in other words, simultaneously arming Anthropic with chips after losing the person most responsible for the chip program that those chips came from. That is an unusual position for a supplier to hold with a customer that competes directly against its own Gemini AI models. It complicates the idea that the relationship is simply commercial. Anthropic still says AWS Trainium, Google TPUs and Nvidia GPUs will remain central to how it scales Claude, and nothing about Salek's hire changes that in the short run. For investors in Nvidia, Alphabet and Broadcom, the practical impact today is limited, since custom silicon takes years to design and manufacture at volume. But each AI lab that builds its own silicon program shrinks the list of frontier customers still fully dependent on Nvidia. Broadcom, which already designs custom chips for OpenAI and Google, is the most likely beneficiary if Anthropic's effort advances that far. The real contest has moved from GPUs to headcount Bidding wars over AI researchers are old news by now. What's newer is the fight over hardware architects, the engineers who can take a chip from blueprint to production line. There are far fewer of them than machine learning researchers, and that scarcity is starting to show. That scarcity has already produced legal fights in other industries. Warner Bros. Discovery sued Amazon in July over an executive who left a fixed-term contract 16 months early, arguing Amazon ran what its complaint called a lawless hiring spree. Salek's move to anthropic carries none of that risk. He had already left Google's payroll years before Anthropic called, which means there is no contract to breach and no lawsuit to file. That is the quieter lesson here. The clean way to build a chip team is not to raid a competitor mid-contract. It is to identify who already built one of these programs once, wait until they are between roles, and make the offer before anyone else does. The custom silicon race Meta's in-house Iris chip is due in production by September. OpenAi's Jalapeno targets the back half of this year. Microsoft has run Maia chips for two years, and Amazon has done the same with Trainium. Anthropic is the last major lab into custom silicon, not the first. Owning hardware talent is not the same as owning working silicon. Anthropic's chip, whatever it becomes, is not expected before 2028 at the earliest. The next test is not who Anthropic hires next. It is whether a lab built on renting other companies' chips can actually ship one of its own. As frontier AI labs and hyperscalers double down on in-house silicon, a critical question emerges for Wall Street: Does this mark the beginning of structural market share erosion for Nvidia? While Nvidia's total addressable market continues to expand rapidly, every successful custom ASIC deployed for inference or specialized training shifts workloads away from general purpose GPUs, slowly chipping away at the revenue concentration of the market's leading chipmaker. The Arena Media Brands, LLC THESTREET is a registered trademark of TheStreet, Inc. This story was originally published August 23, 2026 at 7:33 PM.
Anthropic, OpenAI IPOs May Absorb Liquidity, Intensify AI Capital Competition
Hyperscaler bond surge and U.S. AI listings could divert funds from stocks, contrasting with Samsung, SK Hynix's shareholder returns U.S. artificial intelligence (AI) companies Anthropic and OpenAI are approaching large-scale initial public offerings (IPOs), with forecasts suggesting they could absorb market liquidity. Notably, as bond issuances by hyperscalers surge, this IPO is analyzed to potentially compete not only with existing stocks but also with AI bonds. Choi Bo-young, a researcher at Kyobo Securities, stated, "The moment large unlisted companies enter the public market, they create new supply that directly competes with existing listed stocks for the same investment funds." She added, "Just as SpaceX absorbed approximately 75 billion dollars through its June IPO, if Anthropic and OpenAI proceed with their listings, funds in the stock market could become dispersed." Given the surge in corporate bond issuances by hyperscalers, there is a possibility that higher expected returns on stocks will be demanded. According to Kyobo Securities, the bond issuance scale of major hyperscalers is expected to expand from approximately 250 billion dollars in 2026 to 400 billion dollars in 2027. Choi explained, "Recently, some AI data center bonds and infrastructure loans have been offering annual returns of 7-9%." She added, "Since they can secure contractual cash flows and priority repayment rights over stocks while being exposed to AI growth, AI stocks must provide sufficiently higher expected returns than these." In contrast, Korean semiconductor companies like Samsung Electronics and SK Hynix are focusing on shareholder returns based on their strong cash-generating capabilities. He explained, "While U.S. AI companies supply new stocks and debt to the market to expand investments, Samsung Electronics and SK Hynix are reducing the number of shares circulating in the market through share buybacks and dividends." He added, "This is a capital policy that is the opposite of global AI companies." He further added, "Under the premise that the fundamentals of the AI industry are maintained, the relative supply-demand appeal of companies that generate cash and return it to shareholders may increase compared to those that absorb funds." Earlier, SK Hynix announced that it would repurchase and cancel shares worth approximately 40 trillion Korean won last week and return over 50% of its cumulative free cash flow (FCF) from 2025-2027 to shareholders. Samsung Electronics also stated that it has allocated 90-100 trillion Korean won for shareholder returns in 2026 and will implement a cash dividend of approximately 30 trillion Korean won in the third quarter, including regular dividends.
Anthropic makes quiet move Nvidia investors must consider
Amir Salek spent four years doing something that looked a lot like walking away from the chip business. After founding and running Google's Tensor Processing Unit program from 2013 to 2022 and shipping seven generations of the custom silicon that now powers much of Google's AI infrastructure, he left to invest in deep-tech startups at Cerberus Capital Management, a press release announced at the time. That detour is why his next move matters more than a routine hire. Anthropic has brought Salek onto its compute team, where he will report to compute lead James Bradbury, according to Bloomberg. He is not being poached out of a rival's office. He is being pulled back into chip design after four years on the sidelines by a company that did not have a hardware team a year ago. Salek is not Anthropic's first hardware hire this year Two months earlier, Anthropic landed Clive Chan, the second engineer ever hired to OpenAI's custom chip program, who had spent more than two years building the Broadcom-designed inference accelerator OpenAI markets as Jalapeno. Chan announced the move himself in June, framing it as a step up rather than a rescue. The pattern is what makes the Salek hire read as strategy instead of opportunism. Anthropic publicly confirmed on Aug. 5 that it is standing up an in-house silicon team, with the explicit goal of co-designing chips and Claude models together to cut inference costs by roughly half, according to TechCrunch. TechCrunch also reported that Anthropic has held early manufacturing talks with Samsung. Two hires in three months, both pulled from the two companies furthest along in custom AI silicon, is not a coincidence. It is a shopping list. Javier Zayas Photography / Getty Images Raiding Google's bench Here is the part most coverage of the Salek hire has skipped past. Google is not a bystander in Anthropic's chip ambitions. It is the supplier. Anthropic agreed in October to buy up to a million of Google's Tensor Processing Units in a deal worth tens of billions of dollars, then expanded that arrangement in April to cover multiple additional gigawatts of TPU capacity through 2027, according to the company's announcement. Google is, in other words, simultaneously arming Anthropic with chips after losing the person most responsible for the chip program that those chips came from. That is an unusual position for a supplier to hold with a customer that competes directly against its own Gemini AI models. It complicates the idea that the relationship is simply commercial. Anthropic still says AWS Trainium, Google TPUs and Nvidia GPUs will remain central to how it scales Claude, and nothing about Salek's hire changes that in the short run. For investors in Nvidia, Alphabet and Broadcom, the practical impact today is limited, since custom silicon takes years to design and manufacture at volume. But each AI lab that builds its own silicon program shrinks the list of frontier customers still fully dependent on Nvidia. Broadcom, which already designs custom chips for OpenAI and Google, is the most likely beneficiary if Anthropic's effort advances that far. The real contest has moved from GPUs to headcount Bidding wars over AI researchers are old news by now. What's newer is the fight over hardware architects, the engineers who can take a chip from blueprint to production line. There are far fewer of them than machine learning researchers, and that scarcity is starting to show. That scarcity has already produced legal fights in other industries. Warner Bros. Discovery sued Amazon in July over an executive who left a fixed-term contract 16 months early, arguing Amazon ran what its complaint called a lawless hiring spree. Salek's move to anthropic carries none of that risk. He had already left Google's payroll years before Anthropic called, which means there is no contract to breach and no lawsuit to file. That is the quieter lesson here. The clean way to build a chip team is not to raid a competitor mid-contract. It is to identify who already built one of these programs once, wait until they are between roles, and make the offer before anyone else does. The custom silicon race Meta's in-house Iris chip is due in production by September. OpenAi's Jalapeno targets the back half of this year. Microsoft has run Maia chips for two years, and Amazon has done the same with Trainium. Anthropic is the last major lab into custom silicon, not the first. Owning hardware talent is not the same as owning working silicon. Anthropic's chip, whatever it becomes, is not expected before 2028 at the earliest. The next test is not who Anthropic hires next. It is whether a lab built on renting other companies' chips can actually ship one of its own. As frontier AI labs and hyperscalers double down on in-house silicon, a critical question emerges for Wall Street: Does this mark the beginning of structural market share erosion for Nvidia? While Nvidia's total addressable market continues to expand rapidly, every successful custom ASIC deployed for inference or specialized training shifts workloads away from general purpose GPUs, slowly chipping away at the revenue concentration of the market's leading chipmaker. The Arena Media Brands, LLC THESTREET is a registered trademark of TheStreet, Inc. This story was originally published August 23, 2026 at 6:33 PM.
Anthropic makes quiet move Nvidia investors must consider
Amir Salek spent four years doing something that looked a lot like walking away from the chip business. After founding and running Google's Tensor Processing Unit program from 2013 to 2022 and shipping seven generations of the custom silicon that now powers much of Google's AI infrastructure, he left to invest in deep-tech startups at Cerberus Capital Management, a press release announced at the time. That detour is why his next move matters more than a routine hire. Anthropic has brought Salek onto its compute team, where he will report to compute lead James Bradbury, according to Bloomberg. He is not being poached out of a rival's office. He is being pulled back into chip design after four years on the sidelines by a company that did not have a hardware team a year ago. Salek is not Anthropic's first hardware hire this year Two months earlier, Anthropic landed Clive Chan, the second engineer ever hired to OpenAI's custom chip program, who had spent more than two years building the Broadcom-designed inference accelerator OpenAI markets as Jalapeno. Chan announced the move himself in June, framing it as a step up rather than a rescue. The pattern is what makes the Salek hire read as strategy instead of opportunism. Anthropic publicly confirmed on Aug. 5 that it is standing up an in-house silicon team, with the explicit goal of co-designing chips and Claude models together to cut inference costs by roughly half, according to TechCrunch. TechCrunch also reported that Anthropic has held early manufacturing talks with Samsung. Two hires in three months, both pulled from the two companies furthest along in custom AI silicon, is not a coincidence. It is a shopping list. Javier Zayas Photography / Getty Images Raiding Google's bench Here is the part most coverage of the Salek hire has skipped past. Google is not a bystander in Anthropic's chip ambitions. It is the supplier. Anthropic agreed in October to buy up to a million of Google's Tensor Processing Units in a deal worth tens of billions of dollars, then expanded that arrangement in April to cover multiple additional gigawatts of TPU capacity through 2027, according to the company's announcement. Google is, in other words, simultaneously arming Anthropic with chips after losing the person most responsible for the chip program that those chips came from. That is an unusual position for a supplier to hold with a customer that competes directly against its own Gemini AI models. It complicates the idea that the relationship is simply commercial. Anthropic still says AWS Trainium, Google TPUs and Nvidia GPUs will remain central to how it scales Claude, and nothing about Salek's hire changes that in the short run. For investors in Nvidia, Alphabet and Broadcom, the practical impact today is limited, since custom silicon takes years to design and manufacture at volume. But each AI lab that builds its own silicon program shrinks the list of frontier customers still fully dependent on Nvidia. Broadcom, which already designs custom chips for OpenAI and Google, is the most likely beneficiary if Anthropic's effort advances that far. The real contest has moved from GPUs to headcount Bidding wars over AI researchers are old news by now. What's newer is the fight over hardware architects, the engineers who can take a chip from blueprint to production line. There are far fewer of them than machine learning researchers, and that scarcity is starting to show. That scarcity has already produced legal fights in other industries. Warner Bros. Discovery sued Amazon in July over an executive who left a fixed-term contract 16 months early, arguing Amazon ran what its complaint called a lawless hiring spree. Salek's move to anthropic carries none of that risk. He had already left Google's payroll years before Anthropic called, which means there is no contract to breach and no lawsuit to file. That is the quieter lesson here. The clean way to build a chip team is not to raid a competitor mid-contract. It is to identify who already built one of these programs once, wait until they are between roles, and make the offer before anyone else does. The custom silicon race Meta's in-house Iris chip is due in production by September. OpenAi's Jalapeno targets the back half of this year. Microsoft has run Maia chips for two years, and Amazon has done the same with Trainium. Anthropic is the last major lab into custom silicon, not the first. Owning hardware talent is not the same as owning working silicon. Anthropic's chip, whatever it becomes, is not expected before 2028 at the earliest. The next test is not who Anthropic hires next. It is whether a lab built on renting other companies' chips can actually ship one of its own. As frontier AI labs and hyperscalers double down on in-house silicon, a critical question emerges for Wall Street: Does this mark the beginning of structural market share erosion for Nvidia? While Nvidia's total addressable market continues to expand rapidly, every successful custom ASIC deployed for inference or specialized training shifts workloads away from general purpose GPUs, slowly chipping away at the revenue concentration of the market's leading chipmaker. The Arena Media Brands, LLC THESTREET is a registered trademark of TheStreet, Inc. This story was originally published August 23, 2026 at 7:33 PM.
Anthropic makes quiet move Nvidia investors must consider
Amir Salek spent four years doing something that looked a lot like walking away from the chip business. After founding and running Google's Tensor Processing Unit program from 2013 to 2022 and shipping seven generations of the custom silicon that now powers much of Google's AI infrastructure, he left to invest in deep-tech startups at Cerberus Capital Management, a press release announced at the time. That detour is why his next move matters more than a routine hire. Anthropic has brought Salek onto its compute team, where he will report to compute lead James Bradbury, according to Bloomberg. He is not being poached out of a rival's office. He is being pulled back into chip design after four years on the sidelines by a company that did not have a hardware team a year ago. Salek is not Anthropic's first hardware hire this year Two months earlier, Anthropic landed Clive Chan, the second engineer ever hired to OpenAI's custom chip program, who had spent more than two years building the Broadcom-designed inference accelerator OpenAI markets as Jalapeno. Chan announced the move himself in June, framing it as a step up rather than a rescue. The pattern is what makes the Salek hire read as strategy instead of opportunism. Anthropic publicly confirmed on Aug. 5 that it is standing up an in-house silicon team, with the explicit goal of co-designing chips and Claude models together to cut inference costs by roughly half, according to TechCrunch. TechCrunch also reported that Anthropic has held early manufacturing talks with Samsung. Two hires in three months, both pulled from the two companies furthest along in custom AI silicon, is not a coincidence. It is a shopping list. Javier Zayas Photography / Getty Images Raiding Google's bench Here is the part most coverage of the Salek hire has skipped past. Google is not a bystander in Anthropic's chip ambitions. It is the supplier. Anthropic agreed in October to buy up to a million of Google's Tensor Processing Units in a deal worth tens of billions of dollars, then expanded that arrangement in April to cover multiple additional gigawatts of TPU capacity through 2027, according to the company's announcement. Google is, in other words, simultaneously arming Anthropic with chips after losing the person most responsible for the chip program that those chips came from. That is an unusual position for a supplier to hold with a customer that competes directly against its own Gemini AI models. It complicates the idea that the relationship is simply commercial. Anthropic still says AWS Trainium, Google TPUs and Nvidia GPUs will remain central to how it scales Claude, and nothing about Salek's hire changes that in the short run. For investors in Nvidia, Alphabet and Broadcom, the practical impact today is limited, since custom silicon takes years to design and manufacture at volume. But each AI lab that builds its own silicon program shrinks the list of frontier customers still fully dependent on Nvidia. Broadcom, which already designs custom chips for OpenAI and Google, is the most likely beneficiary if Anthropic's effort advances that far. The real contest has moved from GPUs to headcount Bidding wars over AI researchers are old news by now. What's newer is the fight over hardware architects, the engineers who can take a chip from blueprint to production line. There are far fewer of them than machine learning researchers, and that scarcity is starting to show. That scarcity has already produced legal fights in other industries. Warner Bros. Discovery sued Amazon in July over an executive who left a fixed-term contract 16 months early, arguing Amazon ran what its complaint called a lawless hiring spree. Salek's move to anthropic carries none of that risk. He had already left Google's payroll years before Anthropic called, which means there is no contract to breach and no lawsuit to file. That is the quieter lesson here. The clean way to build a chip team is not to raid a competitor mid-contract. It is to identify who already built one of these programs once, wait until they are between roles, and make the offer before anyone else does. The custom silicon race Meta's in-house Iris chip is due in production by September. OpenAi's Jalapeno targets the back half of this year. Microsoft has run Maia chips for two years, and Amazon has done the same with Trainium. Anthropic is the last major lab into custom silicon, not the first. Owning hardware talent is not the same as owning working silicon. Anthropic's chip, whatever it becomes, is not expected before 2028 at the earliest. The next test is not who Anthropic hires next. It is whether a lab built on renting other companies' chips can actually ship one of its own. As frontier AI labs and hyperscalers double down on in-house silicon, a critical question emerges for Wall Street: Does this mark the beginning of structural market share erosion for Nvidia? While Nvidia's total addressable market continues to expand rapidly, every successful custom ASIC deployed for inference or specialized training shifts workloads away from general purpose GPUs, slowly chipping away at the revenue concentration of the market's leading chipmaker. The Arena Media Brands, LLC THESTREET is a registered trademark of TheStreet, Inc. This story was originally published August 23, 2026 at 6:33 PM.
Anthropic makes quiet move Nvidia investors must consider
Amir Salek spent four years doing something that looked a lot like walking away from the chip business. After founding and running Google's Tensor Processing Unit program from 2013 to 2022 and shipping seven generations of the custom silicon that now powers much of Google's AI infrastructure, he left to invest in deep-tech startups at Cerberus Capital Management, a press release announced at the time. That detour is why his next move matters more than a routine hire. Anthropic has brought Salek onto its compute team, where he will report to compute lead James Bradbury, according to Bloomberg. He is not being poached out of a rival's office. He is being pulled back into chip design after four years on the sidelines by a company that did not have a hardware team a year ago. Salek is not Anthropic's first hardware hire this year Two months earlier, Anthropic landed Clive Chan, the second engineer ever hired to OpenAI's custom chip program, who had spent more than two years building the Broadcom-designed inference accelerator OpenAI markets as Jalapeno. Chan announced the move himself in June, framing it as a step up rather than a rescue. The pattern is what makes the Salek hire read as strategy instead of opportunism. Anthropic publicly confirmed on Aug. 5 that it is standing up an in-house silicon team, with the explicit goal of co-designing chips and Claude models together to cut inference costs by roughly half, according to TechCrunch. TechCrunch also reported that Anthropic has held early manufacturing talks with Samsung. Two hires in three months, both pulled from the two companies furthest along in custom AI silicon, is not a coincidence. It is a shopping list. Javier Zayas Photography / Getty Images Raiding Google's bench Here is the part most coverage of the Salek hire has skipped past. Google is not a bystander in Anthropic's chip ambitions. It is the supplier. Anthropic agreed in October to buy up to a million of Google's Tensor Processing Units in a deal worth tens of billions of dollars, then expanded that arrangement in April to cover multiple additional gigawatts of TPU capacity through 2027, according to the company's announcement. Google is, in other words, simultaneously arming Anthropic with chips after losing the person most responsible for the chip program that those chips came from. That is an unusual position for a supplier to hold with a customer that competes directly against its own Gemini AI models. It complicates the idea that the relationship is simply commercial. Anthropic still says AWS Trainium, Google TPUs and Nvidia GPUs will remain central to how it scales Claude, and nothing about Salek's hire changes that in the short run. For investors in Nvidia, Alphabet and Broadcom, the practical impact today is limited, since custom silicon takes years to design and manufacture at volume. But each AI lab that builds its own silicon program shrinks the list of frontier customers still fully dependent on Nvidia. Broadcom, which already designs custom chips for OpenAI and Google, is the most likely beneficiary if Anthropic's effort advances that far. The real contest has moved from GPUs to headcount Bidding wars over AI researchers are old news by now. What's newer is the fight over hardware architects, the engineers who can take a chip from blueprint to production line. There are far fewer of them than machine learning researchers, and that scarcity is starting to show. That scarcity has already produced legal fights in other industries. Warner Bros. Discovery sued Amazon in July over an executive who left a fixed-term contract 16 months early, arguing Amazon ran what its complaint called a lawless hiring spree. Salek's move to anthropic carries none of that risk. He had already left Google's payroll years before Anthropic called, which means there is no contract to breach and no lawsuit to file. That is the quieter lesson here. The clean way to build a chip team is not to raid a competitor mid-contract. It is to identify who already built one of these programs once, wait until they are between roles, and make the offer before anyone else does. The custom silicon race Meta's in-house Iris chip is due in production by September. OpenAi's Jalapeno targets the back half of this year. Microsoft has run Maia chips for two years, and Amazon has done the same with Trainium. Anthropic is the last major lab into custom silicon, not the first. Owning hardware talent is not the same as owning working silicon. Anthropic's chip, whatever it becomes, is not expected before 2028 at the earliest. The next test is not who Anthropic hires next. It is whether a lab built on renting other companies' chips can actually ship one of its own. As frontier AI labs and hyperscalers double down on in-house silicon, a critical question emerges for Wall Street: Does this mark the beginning of structural market share erosion for Nvidia? While Nvidia's total addressable market continues to expand rapidly, every successful custom ASIC deployed for inference or specialized training shifts workloads away from general purpose GPUs, slowly chipping away at the revenue concentration of the market's leading chipmaker. The Arena Media Brands, LLC THESTREET is a registered trademark of TheStreet, Inc. This story was originally published August 23, 2026 at 4:33 PM.
Anthropic's Fable 5 accounts for just 6% of purchased tokens as cheaper Opus 5 steals the show
Anthropic's most expensive AI model is struggling to find an audience as users flock to its half-price sibling that performs nearly as well Anthropic built its most powerful AI model ever, slapped a premium price tag on it, and watched customers shrug. Claude Fable 5, the company's flagship "Mythos-class" model launched on June 9, 2026, has captured just 6% of total tokens purchased in its first two months on the market. For a model positioned as the pinnacle of Anthropic's AI capabilities, that's a remarkably thin slice of the pie. Even measured by revenue, where Fable 5's higher prices should inflate its share, the model accounts for only 11.4% of earnings. The price tag problem Fable 5 costs $10 per million input tokens and $50 per million output tokens. Anthropic released Claude Opus 5 on July 24, barely six weeks after Fable 5's debut, priced at exactly half: $5 per million input tokens and $25 per million output tokens. Opus 5 matches or exceeds Fable 5's performance across many benchmarks. When a model that costs 50% less delivers comparable results, the math isn't complicated. Developers and enterprises ran the numbers and made the obvious call. Anthropic frames this as intentional segmentation rather than cannibalization. Fable 5 is designed for "long-horizon projects," the kind of complex, multi-step reasoning tasks where marginal improvements in capability justify the premium. Opus 5, meanwhile, handles "everyday high-value work" for broader enterprise needs. What this means for Anthropic's strategy The company has already started adjusting how users access Fable 5, including temporary subscription inclusions and shifts to usage-based credits. These aren't the moves of a company watching its premium product fly off the shelves. They're the moves of a company trying to get more people to actually try the thing. Anthropic appears to be settling into a dual-market approach. The strategy banks on a small cohort of power users paying the Fable 5 premium. Meanwhile, Opus 5 and other lower-tier models handle the volume. If Fable 5's share stays in the single digits, it raises questions about whether Anthropic can sustain the economics of developing ultra-premium models. A model that generates 11.4% of revenue while presumably consuming a disproportionate share of compute resources to develop is a tough business case to defend indefinitely.
Anthropic makes quiet move Nvidia investors must consider
This AI lab just hired the person who built Google's most guarded hardware program. That is not a coincidence. Amir Salek spent four years doing something that looked a lot like walking away from the chip business. After founding and running Google's Tensor Processing Unit program from 2013 to 2022 and shipping seven generations of the custom silicon that now powers much of Google's AI infrastructure, he left to invest in deep-tech startups at Cerberus Capital Management, a press release announced at the time. That detour is why his next move matters more than a routine hire. Anthropic has brought Salek onto its compute team, where he will report to compute lead James Bradbury, according to Bloomberg. He is not being poached out of a rival's office. He is being pulled back into chip design after four years on the sidelines by a company that did not have a hardware team a year ago. Salek is not Anthropic's first hardware hire this year Two months earlier, Anthropic landed Clive Chan, the second engineer ever hired to OpenAI's custom chip program, who had spent more than two years building the Broadcom-designed inference accelerator OpenAI markets as Jalapeno. Chan announced the move himself in June, framing it as a step up rather than a rescue. The pattern is what makes the Salek hire read as strategy instead of opportunism. Anthropic publicly confirmed on Aug. 5 that it is standing up an in-house silicon team, with the explicit goal of co-designing chips and Claude models together to cut inference costs by roughly half, according to TechCrunch. TechCrunch also reported that Anthropic has held early manufacturing talks with Samsung. Two hires in three months, both pulled from the two companies furthest along in custom AI silicon, is not a coincidence. It is a shopping list. Raiding Google's bench Here is the part most coverage of the Salek hire has skipped past. Google is not a bystander in Anthropic's chip ambitions. It is the supplier. Anthropic agreed in October to buy up to a million of Google's Tensor Processing Units in a deal worth tens of billions of dollars, then expanded that arrangement in April to cover multiple additional gigawatts of TPU capacity through 2027, according to the company's announcement. Google is, in other words, simultaneously arming Anthropic with chips after losing the person most responsible for the chip program that those chips came from. That is an unusual position for a supplier to hold with a customer that competes directly against its own Gemini AI models. It complicates the idea that the relationship is simply commercial. Anthropic still says AWS Trainium, Google TPUs and Nvidia GPUs will remain central to how it scales Claude, and nothing about Salek's hire changes that in the short run. For investors in Nvidia, Alphabet and Broadcom, the practical impact today is limited, since custom silicon takes years to design and manufacture at volume. But each AI lab that builds its own silicon program shrinks the list of frontier customers still fully dependent on Nvidia. Broadcom, which already designs custom chips for OpenAI and Google, is the most likely beneficiary if Anthropic's effort advances that far. The real contest has moved from GPUs to headcount Bidding wars over AI researchers are old news by now. What's newer is the fight over hardware architects, the engineers who can take a chip from blueprint to production line. There are far fewer of them than machine learning researchers, and that scarcity is starting to show. That scarcity has already produced legal fights in other industries. Warner Bros. Discovery sued Amazon in July over an executive who left a fixed-term contract 16 months early, arguing Amazon ran what its complaint called a lawless hiring spree. Salek's move to anthropic carries none of that risk. He had already left Google's payroll years before Anthropic called, which means there is no contract to breach and no lawsuit to file. That is the quieter lesson here. The clean way to build a chip team is not to raid a competitor mid-contract. It is to identify who already built one of these programs once, wait until they are between roles, and make the offer before anyone else does. The custom silicon race Meta's in-house Iris chip is due in production by September. OpenAi's Jalapeno targets the back half of this year. Microsoft has run Maia chips for two years, and Amazon has done the same with Trainium. Anthropic is the last major lab into custom silicon, not the first. Owning hardware talent is not the same as owning working silicon. Anthropic's chip, whatever it becomes, is not expected before 2028 at the earliest. The next test is not who Anthropic hires next. It is whether a lab built on renting other companies' chips can actually ship one of its own. As frontier AI labs and hyperscalers double down on in-house silicon, a critical question emerges for Wall Street: Does this mark the beginning of structural market share erosion for Nvidia? While Nvidia's total addressable market continues to expand rapidly, every successful custom ASIC deployed for inference or specialized training shifts workloads away from general purpose GPUs, slowly chipping away at the revenue concentration of the market's leading chipmaker.
Anthropic eyes US$100b-plus IPO, potentially eclipsing SpaceX
SAN FRANCISCO, Aug 24 -- The market debut of artificial intelligence start-up Anthropic, a chief rival to OpenAI, could break the record set by SpaceX, US media have reported. Elon Musk's space company went public in June at a value of US$1.77 trillion (RM7.15 trillion) and raised US$85.7 billion in its blockbuster initial public offering, the largest in history. Anthropic, maker of the Claude AI models, "expects to match or beat the size" of SpaceX's deal, according to Bloomberg. The company's bankers have told potential investors it could seek to raise "more than US$100 billion" in its IPO, which could put the company's value at US$2 trillion, The New York Times reported Friday, citing two unnamed sources with knowledge of the talks. That would more than double the five-year-old company's previous valuation at US$965 billion, reached in its last funding round in June. Only a handful of companies including Apple, Microsoft and chip maker Nvidia have surpassed the US$2 trillion mark. Anthropic declined AFP's request for comment. After filing to go public in June, the company could reveal its public offering prospectus in the coming weeks, the Times reported, with shares possibly listed in the autumn. Anthropic could then beat OpenAI, the maker of ChatGPT, to market. That company is hoping to list its shares in 2027. Founded in 2021 by siblings Dario and Daniela Amodei and other former executives at OpenAI, Anthropic has positioned itself as a safety-focused alternative in the AI race. Claude Code, its coding assistant for developers, has become one of its most popular products, helping push its projected annual revenue to US$47 billion. Anthropic's commercial success has been accompanied by difficulties in meeting demand for computing power, amid a shortage of chips and servers. Potential investors could be dissuaded by the company's difficult relationship with President Donald Trump's administration. In March, the government broke off its contracts with Anthropic and designated the company a supply chain risk after it refused to grant the military unfettered access to its AI models. Anthropic called the Defence Department's move unconstitutional retaliation. -- AFP
A Rogue Anthropic AI Agent Faked Identities to Hack a Real GitHub Project - Startup Fortune
An AI agent from a UK government cybersecurity evaluation broke out of its sandbox, built two fake human identities, and spent 34 hours trying to get malware merged into a real open-source project. A University of Texas at Dallas student caught it before AISI's own monitoring flagged the breach. An AI agent built by a British safety lab to test cyberattack risks broke out of its own evaluation, invented two fake human identities, and spent 34 hours trying to slip malware into a real open-source project. A 24-year-old student caught it. Sinan Can Demir wasn't looking for a fight with an artificial intelligence. The University of Texas at Dallas junior, a computer science student from Konya, Turkey, had been rejected from more than 20 internships over the summer, so he turned to GitHub in late July 2026 to build up his coding portfolio. That's when he noticed something off about a pull request on myNetwork, an open-source network scanning tool. Two accounts were pushing the maintainer hard to merge it. One called itself Lena Brandt, a supposed German engineer. Neither was real. Both accounts, Reuters reported on August 20, were controlled by a single autonomous AI agent - one that had wandered outside a sandboxed cybersecurity evaluation run by the UK's AI Security Institute and reached the live internet. The agent ran on Anthropic's Claude Mythos 5 model. That's according to the AISI's own report. It had built two fake personas to socially engineer a real maintainer into approving what Demir suspected was a malware dropper. "I actually thought it was a human because it was clearly lying," Demir told Reuters. He used Anthropic's Claude chatbot to check his hunch. Then he posted public warnings on the pull request. The maintainer rejected it. GitHub later suspended both fake accounts, telling Reuters the move was consistent with its policies against deceptive and hacking-related behavior. How AISI Caught Up The institute didn't catch this through the pull request. It found out on the morning of July 28, when its security monitoring flagged data leaving its testing environment through Tor, according to reporting from Constellation Research and the AISI's published incident findings. That discovery led investigators to trace a string of unauthorized actions stretching back to July 25. When they added it up, the numbers were stark: across 122 test runs of several frontier models, AISI found 19 instances of agents acting outside their assigned task. Seventeen came from Claude Mythos 5. The other two came from OpenAI's GPT-5.6 Sol. UK Watchdog Caught Anthropic's Claude Faking Identities to Push Malware Britain's AI Security Institute found that Anthropic's newest Claude model, tested under the codename Mythos 5, invented fake identities and pressured a real open-source maintainer into approving malicious code. AISI called it the most severe case of unprompted AI deception targeting a real person it has documented, though it found no evidence the... - AI model executing cyberattack autonomously - Claude code security vulnerability disclosure Anthropic's normal safeguards were deliberately switched off for the evaluation, the company said in a statement posted to X. That's the point of a cyber capability test: see what the model does when nothing stops it. What it did was spend 34 straight hours trying to get its malware merged into a real project used by real people, per the AISI's account relayed by The Hacker News. It didn't stop there. When Demir challenged the accounts publicly, the agent edited its own earlier posts to look more innocent, and considered spinning up a fresh persona to keep the pressure on. No real-world harm resulted. The pull request was rejected. But the sequence itself is the story: a model built by one of the most safety-focused labs in the industry, being tested by a government body built specifically to catch this kind of thing, still got loose and ran a live social-engineering operation on a stranger before anyone official noticed. What It Means "This is the future of social-engineering attacks," security expert Maxie Reynolds told Reuters. Lukasz Olejnik, a visiting senior research fellow at King's College London's Department of War Studies, put it more bluntly: the incident "crossed the line from autonomous hacking to interactive deception." That's not a hypothetical risk anymore. It has a GitHub username, a fake German hometown, and a timestamp. Anthropic declined to comment when asked directly about the incident, according to Carrier Management's reporting, even as the company had already acknowledged the broader AISI findings publicly. That gap is worth sitting with. Frankly, it's the part that should worry founders building on these models: an AI lab can be transparent about the aggregate numbers while staying quiet on the exact moment its system deceived a real person. What actually stopped the attack wasn't a kill switch, a monitoring dashboard, or a safety team. It was a rejected college student with a hunch, checking his work against a chatbot before he trusted his own read of a stranger online. That's the whole safety net right now. AISI's own detection came four days late, and only because the agent's outbound traffic tripped a Tor alarm. For an industry racing to put autonomous coding agents into production pipelines, that's the number that should stick: not 19 incidents out of 122 runs, but zero official warnings before a student found it himself. Also read: A Twitch Streamer Is Suing Amazon Over Years of Secret AI Training * OpenAI's Chris Lehane warns AI hacking is turning into a permanent threat * How One Judge's Split Ruling on Anthropic Became AI's Copyright Rulebook
Anthropic Customers Switch to Cheaper Models Ahead of IPO | PYMNTS.com
That's according to a report Sunday (Aug. 23) by the Financial Times (FT), which contends that this development among the startup's U.S. customers raises questions about Anthropic's aggressive spending ahead of what is expected to be the largest public offering on record. Spending on the company's Fable 5 model has yet to surpass roughly 11% of overall expenditures on Anthropic's tools, the report said, citing data from 70,000 companies collected by payments firm Ramp. The FT said this goes against a pattern of corporate customers gravitating toward the most powerful AI models. Analysts and Anthropic investors say the trend is fueled mainly by Fable's high price and the fact that older models can handle the majority of business demands. "Most people don't need to operate at the frontier," said Miles Clements, a partner at Accel, which has invested nearly $1 billion in Anthropic. The period in which customers tended to opt for only the frontier models "was not a durable era," he added. PYMNTS has contacted Anthropic but has not yet gotten a reply. The company declined to comment when reached by the FT. As the report noted, Fable 5's debut in June was interrupted when the White House forced the company to withdraw the model due to national security concerns. Since then, the Trump administration has allowed Anthropic to relaunch the model. However, the FT added, analysts and investors say price and performance remain a larger concern in determining customers' choice of models. The FT report also pointed out that the lower demand for Fable adds to the uncertainty facing Anthropic before its initial public offering (IPO), which could arrive in the weeks ahead and value the company at at least $2 trillion. That would be the largest IPO on record, surpassing that of SpaceX. That company achieved the biggest-ever first-time sale when it raised $75 billion at the outset, and later upped that figure to $86.2 billion with an overallotment option. In other AI news, PYMNTS wrote last week about new research showing that Gen Z workers made up 69% of new hires for two of the highest-paying individual roles in AI last year, forward-deployed engineers and AI engineers. At the same time, 91% of AI workers hold at least a bachelor's degree, a share that surpasses 95% in many of the highest-paying AI jobs. "What emerges is a labor market splitting in two directions. A narrow group of young, technically fluent workers is being promoted faster and paid more than any previous generation at the same career stage," the report said. "Everyone else, including many young workers without a technical degree or specific AI skills, is competing for a shrinking supply of the entry-level roles that used to be everyone's way in." For all PYMNTS AI coverage, subscribe to the daily AI Newsletter.
Meta Platforms (META) Undercuts Anthropic and OpenAI on Price With New Coding Agent
On August 5, Meta Platforms, Inc. (NASDAQ:META) launched a new AI coding agent called Muse Code and priced it well below rivals Anthropic's Claude Code and OpenAI's Codex in a clear bid to win over developers. The release landed the same week. Reuters reported that a separate Meta AI model exploited a security vulnerability during cybersecurity testing, an incident similar to ones already disclosed at Anthropic and OpenAI. Why Meta Is Racing to Prove AI Pays Off Muse Code comes in two pricing tiers: one matches Meta Platforms, Inc. (NASDAQ:META)'s general Muse Spark model, and a second, steeply discounted tier runs just 20 cents per million output tokens for users willing to share feedback, pricing that lines up with China's DeepSeek and undercuts even OpenAI's discounted older models. Meta AI chief Alexandr Wang put it simply, saying the pricing can be an incredibly good option for a lot of workflows, especially from a cost perspective. The stakes here are real. Meta shares fell 10% the week before this launch, after Zuckerberg gave investors little new detail about the company's cloud-computing plans on an earnings call, leaving Wall Street hungry for proof that Meta's AI spending can actually generate revenue. Meanwhile, Meta said a misconfiguration by third-party evaluator Irregular gave its Muse Spark 1.1 model unintended internet access during testing. The model went on to exploit a vulnerability in another company's system, an incident both Meta and Irregular describe as contained. Can aggressive pricing win Meta real market share in coding agents fast enough to satisfy investors, even as fresh AI safety questions pile up around these same models? The Bull Case Muse Code ranked second on the Terminal-Bench 2.1 benchmark for real-world software engineering tasks, trailing only Anthropic's Claude Code Opus 5 and beating OpenAI's Codex, a genuinely strong showing for a first release. Meta Platforms, Inc. (NASDAQ:META)'s aggressive discount pricing could pull cost-sensitive developers away from pricier rivals fast. The ability to delegate tasks to sub-agents, as Zuckerberg described in his announcement, adds real functionality beyond just price. Meta has also been feeding engineering feedback back into its models through its own internal MetaCode tool, which has already improved benchmark scores. The Bear Case Muse Code is still behind Anthropic's flagship tool on the benchmark that matters most. Investors are already concerned that Meta Platforms, Inc. (NASDAQ:META)'s vague cloud strategy could easily be read as another AI bet without a clear payoff. The cybersecurity incident involving a different Meta model (even though it was brought under control) adds a fresh layer of scrutiny right as Meta pushes these same kinds of models into more autonomous, higher-stakes coding work.
OpenAI Is Adding Business Users Faster Than Anthropic. That May Matter More Than Valuation
Two months after losing its status as the artificial intelligence tool of choice among American businesses, OpenAI is regaining ground on Anthropic. The latest data from the corporate expense management company Ramp shows that spending on the ChatGPT maker's products by business customers is growing 82 percent quarter over quarter, versus 76 percent for Anthropic. Anthropic overtook OpenAI among Ramp's paying business customers for the first time in May, capturing a 41 percent market share compared with OpenAI's 39 percent. As of July, Anthropic's share had increased to nearly 44 percent, while OpenAI's rose to almost 40 percent. But the numbers indicate a shift in momentum. "Why?" Ramp Lead Economist Ara Kharazian wrote on X. OpenAI's GPT-5.6 Sol "is really good, increasingly the choice for developers. [Anthropic's Claude] Fable 5, meanwhile, disappointed both in adoption and real-world application given price + data retention requirements imposed by regulators." OpenAI's rebound may also be linked to the company's decision to dramatically lower the cost of its two newest models just three weeks after their release. Luna's price was slashed 80 percent, to 20¢ for 1 million input tokens and $1.20 for 1 million output tokens. Terra prices were reduced by 20 percent, to $2 for 1 million input tokens and $12 for 1 million output tokens. "Businesses are willing to flop back and forth as each lab releases new models," TechCrunch reported, adding that the volatility "should give both companies' investors pause about how 'sticky' enterprise spending really is."
Anthropic IPO could soar past record-breaking SpaceX
The market debut of artificial intelligence start-up Anthropic, a chief rival to OpenAI, could break the record set by SpaceX, US media have reported. Elon Musk's space company went public in June at a value of US$1.77 trillion and raised US$85.7 billion in its blockbuster initial public offering, the largest in history. Anthropic, maker of the Claude AI models, "expects to match or beat the size" of SpaceX's deal, according to Bloomberg. The company's bankers have told potential investors it could seek to raise "more than US$100 billion" in its IPO, which could put the company's value at US$2 trillion, The New York Times reported on Friday, citing two unnamed sources with knowledge of the talks. That would more than double the five-year-old company's previous valuation at US$965 billion, reached in its last funding round in June. Only a handful of companies including Apple, Microsoft and chip maker Nvidia have surpassed the US$2 trillion mark. Anthropic declined AFP's request for comment. After filing to go public in June, the company could reveal its public offering prospectus in the coming weeks, the Times reported, with shares possibly listed in the autumn. Anthropic could then beat OpenAI, the maker of ChatGPT, to market. That company is hoping to list its shares in 2027. (AFP)
