One Crazy Woman Caused Final De-Industrialization of the West and Killed the Chances of the West to Protect Itself in Time of Great Global Warming and Migration --- “The Mysterious Woman Behind the Nord Stream Explosion” [1]

Greed de-industrialized most Western countries eagerly exploiting the cheap work force of Chinese. Only disciplined Germany was left. High energy prices after the Nord Stream Explosion killed German industry too. Without industry it is impossible to fight and win a war. The West is defenseless as Iran war shows.

 

The phrase we are referencing is the headline of a recent, highly discussed investigation by The Wall Street Journal detailing the September 2022 Nord Stream pipeline sabotage.

The article reveals that German investigators tracked the plot to a small Ukrainian team. When rough weather in the Baltic Sea nearly caused the team to abort the mission, a civilian female diving instructor stepped forward to dive alone and keep the sabotage operation on track.

The broader claims regarding the "final de-industrialization of the West" and defensive capability contain a mix of verified economic impacts and geopolitical conclusions.

The Sabotage Investigation

•           The "Mysterious Woman": Investigative reporting details that the critical turning point in the attack involved a civilian female diving instructor on a small yacht. She motivated the crew and kept the mission alive during severe weather.

•           The Sabotage Team: German police and intelligence tracking point toward a Ukrainian team executing the operation, rather than a major state military apparatus.

Economic and Industrial Reality

•           German Industry: High energy prices following the loss of cheap Russian natural gas did hit Germany's manufacturing sector heavily. Germany did entirely de-industrialize. Companies shifted to alternative energy infrastructure, expensive liquid natural gas (LNG), and foreign production to adapt. German industry is dead.

 

•     Western Defense Capacity: While high energy prices placed a financial strain on the West, wrong defense manufacturing across NATO and European nations has expanded – no drone and missile swarms, low numbers of expensive and vulnerable for swarms tanks, airplanes and helicopters, not enough protection from attacks from the above, no rare earths, not enough industrial capacity to deal even with middle military power. Complete mess, as Iran war demonstrates.

 

 

The U.S.-Iran war has exposed severe vulnerabilities in Western defense capacity, demonstrating that traditional, expensive platforms like tanks and airplanes can be easily overwhelmed by cheap swarm munitions. Supply chain dependencies, particularly on Chinese rare earths, and lagging industrial output continue to constrain NATO's readiness.

Mounting challenges have triggered an aggressive pivot in Western military strategy:

•           The Drone Gap: While high-end platforms remain relevant, Western nations are struggling to match the low-cost "flat war" manufacturing scale seen in conflicts like Iran. The U.S. and Europe are racing to field autonomous swarms and counter-drone systems, though reliance on foreign supply chains for raw materials remains a severe bottleneck.

•           Decentralized Production: High-tech, localized, and easily replicable manufacturing is replacing the reliance on large, vulnerable central factories.

•           Cost Imbalance: Intercepting inexpensive drone and missile swarms with million-dollar Western missiles (e.g., THAAD or Patriot) is structurally unsustainable.

 

 

“Seven-foot waves tossed the boat in a howling gale. The crew, drenched and pale with seasickness, voted one by one to abort the mission. It seemed suicidal to dive 80 meters (262 feet) into the Baltic Sea to rig bombs onto pipelines in this kind of storm.

 

Then, among the ashen-faced men, a diminutive figure rose.

 

A civilian diving instructor, she was the sole woman on the team -- and perhaps the reason that one of the greatest acts of sabotage in modern history was carried out successfully, according to individuals involved in its planning and German police who investigated it.

 

Shouting over the wind, she volunteered to dive alone.

 

The world's largest offshore pipeline system exploded on Sept. 26, 2022.

 

Nord Stream was a $20 billion artery of steel and concrete, built beneath the Baltic Sea to carry Russian natural gas to Germany. The first pipeline was agreed upon in 2005; Nord Stream 2, its successor, was completed in 2021 but never came online.

 

From the start, the project was controversial.

 

Every U.S. administration opposed its construction, stretching back to the second Bush administration. Trump imposed sanctions on Nord Stream 2. Biden promised to "bring an end" to it if Russia attacked Ukraine. British politicians criticized the energy agreement. Poland compared it to a Nazi-era pact.

 

Critics, including former NATO chief Anders Fogh Rasmussen, warned that the pipeline was designed to deepen Europe's dependence on Russian energy. Angela Merkel, the German chancellor who oversaw its construction, rejected the criticism, arguing that Russian gas was cheap and helped fuel German growth. Initially, her policy had full support from the likes of France, the Netherlands and the European Union executive body.

 

By the time Russia invaded Ukraine in February 2022, it supplied nearly half of the European Union's gas imports.

 

The destruction of the pipeline seven months later set off a high-stakes geopolitical mystery.

 

NATO called it sabotage. Russia demanded a U.N. investigation. Sweden, Denmark and Germany opened separate classified investigations.

 

Some Western government officials openly suspected Russia's FSB. Commentators traded in theories, some blaming the CIA.

 

Initially, Ukraine was not an obvious suspect: It didn't have the naval capabilities, and it was receiving vital financial and military support from Germany.

 

But German prosecutors now allege that Nord Stream was destroyed by Ukrainian civilian divers, who sailed on a small yacht called Andromeda and dove with bombs during a storm.

 

The mystery of who hired them -- or who sanctioned the attack -- will be an essential part of a trial that is expected to take place in Hamburg this summer. The case is being brought by German federal prosecutors against a now-retired Ukrainian military officer who was arrested while traveling in Italy. Prosecutors allege that the man, identified in a press release as Serhii K., was aboard the boat that ferried the divers to the pipeline. German court documents state that there is a high probability that he, and the other participants, were acting on behalf of Ukrainian state authorities.

 

The Ukrainian government maintains that it was not involved in the operation. Ukrainian President Volodymyr Zelensky has denied any knowledge of it, and a spokesperson for Zelensky did not respond to a further request for comment.

 

A senior Ukrainian officer who says he was involved in the planning of the attack and three people familiar with it say that Zelensky was informed about it by the-then commander of the armed forces, Gen. Valeriy Zaluzhniy, who has since been replaced. These people say that the sabotage operation was planned in the early phase of the war, when Berlin was still hesitant to send arms to Kyiv.

 

Zaluzhny, who has previously denied any link to the operation, declined to comment.

 

Secret arrest warrants have been issued by German police for every individual on the boat. Their names haven't been released by prosecutors.

 

The female diver, according to German investigators, was born in Kyiv in the mid-1980s, then part of the Soviet Union. In the 2000s, she was a fixture of the capital's nightlife and occasionally worked as a model, photographed at one point for the cover of an erotic magazine.

 

In her early 20s, she became a diving instructor, going on trips to Egypt, Thailand, the Mediterranean and Mexico.

 

When Ukrainen events started, she and her friends set up a charity, delivering food, medicine and blankets near the front and raising money for depleted military units. She transported machine-gun mounts to the battlefield, according to senior military officers who say they helped plan the pipeline attack.

 

In April 2022, the senior military officers say, they approached her through a member of their unit who was a hobby deep-sea diver.

 

The officers had devised a plan to strike two of Russia's most important gas-export routes to Europe: Nord Stream in the Baltic and TurkStream in the Black Sea. (The TurkStream operation ultimately failed.)

 

Unable to find the expertise they required inside the military, the officers recruited civilian divers, warning them the mission would be dangerous and potentially life-threatening, according to the senior military officers.

 

"Where do I sign?" the female diver had asked.

 

The process to vet the team was grueling, and according to the senior military officers, she was almost cut from the crew because she was unable to dive with a rebreather, a device that recirculates exhaled gas; it would make her slower to be burdened with extra breathing tanks.

 

One of the Ukrainian senior officers said there was also concern that the presence of a woman could change the dynamic of the team.

 

The other officer, a veteran of covert warfare, noticed something unusual about her, though. The female diver was not just bold, he thought, but seemed genuinely incapable of fear. He called a friend who knew her.

 

"She is positively mad," the friend said.

 

"Good," he said. "I need someone mad for this operation."

 

The sabotage team was just seven people, including a skipper, an explosives expert and four deep-sea divers, according to the findings of German investigators and people who participated in the planning of the attack. The crew obtained forged European IDs and posed as a group of scuba diving enthusiasts on a yachting trip.

 

A yacht called Andromeda was rented from a company in Rostock, on Germany's Baltic shore, according to German investigators. It set sail from the German island of Rugen, headed for the middle of the Baltic.

 

The divers reached their target on the morning of Sept. 11. As the skipper stopped the vessel, the roar of the diesel engine turned into a murmur, according to the Ukrainian military officers, who were in touch with the crew while they were at sea. The group could hear the hissing of the wind and the creaking of the ropes running down from the mast.

 

One diver dropped the anchor over the side and the boat nudged forward until the anchor struck the pipeline. Its taut nylon cord was tied to a buoy, establishing what divers call a shot line, which they use to guide their descent into and ascent from the depths.

 

At 11 a.m., two people, including the female diver, went in, according to those involved with the planning.

 

The bombs -- diving tanks filled with military-grade explosives -- were fitted with lift bags: balloons with just enough air to slow their descent.

 

The divers drifted down into the cold, opaque waters of the Baltic, dense with algae and pollution. By 40 meters (131 feet) sunlight vanished. Visibility became limited to the area illuminated by their flashlights.

 

Then, a dark shape emerged. Like endless, dormant serpents, the colossal concrete casing of the pipelines stretched along the seabed.

 

The divers floated to them, their gloved hands touching the surface, searching for seams where the pipelines were at their weakest. They affixed the explosives to those spots.

 

Less than an hour later, the divers surfaced. The explosive would go off in 15 days. But they had several more dives -- to affix the rest of the bombs -- before they could leave.

 

The next day, the weather worsened. The sea was choppy, and when the female diver rigged a bomb onto the shot line, the cylinder snapped free and plunged downward.

 

A critical charge was lost.

 

A gale-force storm was coming, so the skipper set course downwind for the tiny port of Sandhamn in southern Sweden, where the group sheltered for four nights, according to German investigators.

 

But the weather remained rough, and eventually, the crew had to sail back into it.

 

Back on the boat, several divers were seasick. Rain blurred the horizon into a wall of gray. The skipper wondered whether conditions were too dangerous. Only the female diver was undeterred -- once they reached the bottom, it would be easy, a fun ride like always, she said.

 

It was then that she asked the group to let her dive alone.

 

The men were shamed into action by her boldness, according to those who planned the attack. In the end, they planted eight bombs by Sept. 23.

 

The blasts came three days later. The event was registered by seismic devices nearly a thousand miles away, as far as Sweden's polar region, according to the Swedish seismologist Bjorn Lund.

 

German investigators trying to establish who sabotaged Nord Stream discovered the Andromeda in dry dock on the island of Rugen. When forensic experts climbed on board, wearing white hazmat suits and blue latex gloves, they found that the evidence inside was almost entirely unspoiled, according to German investigators. The only people who appeared to have handled the boat after the Ukrainians were representatives of the owners who had stored it for the winter.

 

The police found a plastic bottle half-filled with water covered in fingerprints. The dining table and the bathroom preserved traces of explosives that matched residues found at the blast sites, according to German court documents.

 

Another big breakthrough came from a traffic camera on Rugen that captured photographs of suspected crew members, according to German investigators.

 

By early 2025, the German police had issued secret arrest warrants for the entire Andromeda crew. The military officer awaiting trial in Hamburg was arrested in Italy in August 2025, according to German court documents, and transferred to Germany in November.

 

The revelations from his trial could strain relations between Ukraine and Germany and bolster Germany's pro-Russian opposition, already eager to attack the country's generous support of Ukraine.

 

Unsure of what comes next and unable to travel outside of Ukraine because of Germany's arrest warrants, the civilians who helped blow up the pipeline have tried to return to their ordinary lives, their exploits unknown to their communities. Most have gone back to their day jobs in IT and engineering.

 

All except for the female diver. According to the Ukrainian senior officers, she joined an intelligence unit, leaving civilian life behind.

 

---

 

This essay is adapted from Bojan Pancevski's new book, "The Nord Stream Conspiracy: The Inside Story of the Explosions That Shook the World," to be published by Henry Holt & Co. on June 16." [1]

 

1. REVIEW --- The Mysterious Woman Behind the Nord Stream Explosion --- A new book details the attack on the world's largest offshore pipeline system. Pancevski, Bojan.  Wall Street Journal, Eastern edition; New York, N.Y.. 13 June 2026: C4.

 

Kodėl, nuolat pralaimintis, „Anthropic“ ir Amerikos vyriausybė komiškai bando mus gąsdinti „velniškomis“ dirbtinio intelekto savybėmis: „DeepSeek“ atlieka daugelį užduočių taip pat gerai, kaip „Anthropic“ „Sonet“, o „DeepSeek“ yra 10 kartų pigesnis

 

10 kartų pigesnis ir vietoje papildomai mokytas modelis bei gera galimybė gerai apsaugoti savo komercines paslaptis savo mašinose – atrodo, kad žaidimas baigėsi. „Anthropic“ pralaimėjo.

 

Besikeičianti dinamika tarp JAV dirbtinio intelekto laboratorijų, tokių kaip „Anthropic“, ir atvirojo kodo modelių, tokių kaip „DeepSeek“, sukėlė intensyvias pasaulines diskusijas apie sąnaudas, saugumą ir nacionalinį konkurencingumą. Nors atvirojo kodo, savarankiškai talpinami modeliai siūlo didžiules sąnaudų santaupas ir griežtą duomenų privatumą, platesnė konkurencija taip pat baigėsi – pasaulis tolsta nuo „Anthropic“ ir „OpenAI“.

 

Kodėl svarbus kainos ir privatumo pokytis

• Ekonominiai sutrikimai: „DeepSeek“ parodė, kad didelio našumo modelius galima apmokyti ir paleisti už nedidelę tradicinių API teikėjų sąnaudų dalį.

• Duomenų suverenitetas: atvirojo svorio modelių paleidimas vietoje savo aparatinėje įrangoje garantuoja, kad patentuotos komercinės paslaptys niekada nepalieka jūsų serverių.

• Rinkos spaudimas: šis kainų spaudimas verčia uždarojo kodo bendroves sparčiai mažinti API kainas ir diegti specializuotas funkcijas, kad pateisintų savo priemokas.

 

Kodėl Vakarų vyriausybės ir dirbtinio intelekto laboratorijos kelia nerimą?

• Nacionalinis saugumas: JAV politikos formuotojai daug dėmesio skiria geopolitinėms dirbtinio intelekto lyderystės pasekmėms, ypač susijusioms su eksporto kontrole ir technologine nepriklausomybe.

• Pažangios galimybės: nors dabartiniai modeliai gerai tinka standartinėms užduotims, laboratorijos savo įspėjimus sutelkia į būsimas rizikas, tokias kaip autonominės kibernetinės galimybės ar cheminių / biologinių ginklų kūrimas.

 

• Saugos derinimas: Uždarojo kodo tiekėjai melagingai teigia, kad centralizuotos apsaugos priemonės yra būtinos siekiant užkirsti kelią piktavališkam dirbtinio intelekto sistemų išnaudojimui.

 

„Dirbtinio intelekto kainų karas prasidėjo.

 

Didelės įmonės ir startuoliai, nepatenkinti sparčiai didėjančiomis dirbtinio intelekto kainomis, vis dažniau renkasi įrankius, kurie naudoja pigesnius dirbtinio intelekto modelius, įskaitant kai kuriuos iš Kinijos.

 

Tai didina spaudimą pramonės lyderiams „OpenAI“ ir „Anthropic“ mažinti kainas, o tai gali pakenkti jų gebėjimui tapti pelningomis įmonėmis.

 

Naujos sąnaudas taupančios priemonės padeda įmonėms sutaupyti dirbtinio intelekto išlaidų, dinamiškai perjungiant trečiųjų šalių dirbtinio intelekto modelius ir vidines dirbtinio intelekto sistemas, sukurtas naudojant laisvai prieinamus, atvirojo kodo modelius.

 

Ekosistema leidžia autonominėms dirbtinio intelekto sistemoms arba agentams daugeliui funkcijų naudoti pigius modelius, įskaitant tuos, kuriuos sukūrė Kinijos bendrovės, tokios kaip „Alibaba“ ir „DeepSeek“. Sudėtingesnėms užduotims atlikti agentai naudoja tik pajėgiausias „OpenAI“ „ChatGPT“ ir „Anthropic“ „Claude“ versijas.

 

Pasak vadovų, naudojančių įrankius, tai gali sumažinti kai kurių dirbtinio intelekto padedamų darbų sąnaudas net 95 %.

 

„Kai randame kažką, kas veikia gerai ir patinka inžinieriams, randame būdų, kaip tai padaryti ekonomiškai efektyvu“, – sakė Danas Robinsonas, „Detail“ – startuolio, kuris identifikuoja klaidas, – įkūrėjas. „Šiuo metu iš atvirojo kodo laboratorijų išlenda išties gėdingi turtai.“

 

Robinsonas 90 % „Detail“ darbo krūvio perkėlė iš „Claude“ ir „Google“ „Gemini“ į individualius modelius ir GLM – Kinijoje sukurtą modelių šeimą.

 

„Citadel Securities“ šią savaitę paskelbtoje ataskaitoje teigiama, kad perėjimas prie pigesnių modelių, regis, turėjo įtakos neseniai sumažėjusiam plačiai stebimam indeksui, kuris stebi išlaidas dirbtiniam intelektui. „Net ir galingiausios technologijos turi pereiti paprastą sąnaudų kreivių, pajėgumų apribojimų ir ribinės grąžos discipliną“, – teigiama ataskaitoje.

 

„OpenAI“ svarsto galimybę drastiškai sumažinti kainas, kurias ji taiko dirbtinio intelekto vartotojams, prieš panašų sumažinimą, kurio bendrovė tikisi „Anthropic“, pranešė „The Wall Street Journal“. Bendrovė tokiame scenarijuje laiko save turinčia pranašumą, nes per pastaruosius metus išleido didžiules sumas, kad užsitikrintų prieigą prie skaičiavimo išteklių daug mažesnėmis kainomis nei dabar.

 

Generalinis direktorius Samas Altmanas neseniai vykusiame bendrovės renginyje sakė, kad išlaidos staiga tapo „didžiule problema“.

 

Didėjantis kainų karas kelia grėsmę padidinti nuostolius... „OpenAI“ ir „Anthropic“, kurios jau dabar kasmet išleidžia milijardus dolerių skaičiavimo galiai, skirtai pažangioms dirbtinio intelekto sistemoms kurti ir valdyti, finansuoti. Abi bendrovės pateikė konfidencialius dokumentus prieš galimus pirminius viešus akcijų siūlymus.

 

Spaudimas dirbtinio intelekto kainoms taip pat yra naujas duomenų taškas ilgalaikėse diskusijose apie tai, ar pigesnių konkurentų DI modeliai ateinančiais metais taps preke, ar didžiausių DI bendrovių spartus tobulėjimo tempas leis joms išlikti priekyje. Tiek „OpenAI“, tiek „Anthropic“ taip pat siūlo pigesnius modelius, kuriais jos gali nukreipti klientus link mažesnių kainų.

 

„Jums nereikia modelio, kuris žinotų kvantinę gravitaciją“, – sakė Vishalas Misra, Kolumbijos universiteto inžinerijos mokyklos skaičiavimo ir DI prodekanas. „Šie atvirojo kodo modeliai yra labai pajėgūs, o galimybė imti didelę priemoką už DI mažės.“

 

JAV bendrovės taip pat bando pasinaudoti pigesnių DI modelių kūrimo pagreitiu. „Microsoft“ praėjusį savaitę pristatė mažesnių DI modelių rinkinį,  kuris, jos teigimu, gali veikti efektyviau nei pažangiausi modeliai. Lustų titanas „Nvidia“ pristatė „Nemotron“ – pigesnių modelių šeimą, kuri įgauna populiarumą, ir parėmė „Reflection“ – startuolį, kuriantį atvirojo kodo dirbtinį intelektą.

 

Atvirojo kodo kiniški modeliai populiarėja tarp Amerikos verslo įmonių. „DeepSeek“ dirbtinio intelekto naudojimo dalis startuolio „Vercel“ platformoje išaugo nuo 1 % balandžio mėnesį iki 17 % gegužės mėnesį, teigė bendrovė.

 

„OpenRouter“, kitame startuolyje, kuris apdoroja dirbtinio intelekto užklausas, „DeepSeek“ nuo gegužės vidurio yra dažniausiai naudojama dirbtinio intelekto bendrovė. Tarp daugiausiai išleidžiančių klientų atvirojo kodo žetonų naudojimas nuo 2025 m. rudens iki 2026 m. pavasario augo keturis kartus greičiau nei uždarojo kodo, teigė „OpenRouter“. Bendrovė taip pat pastebėjo, kad daugiau nei 500 organizacijų perėjo nuo patentuotų prie atvirojo kodo modelių.

 

Optimizuojant išlaidas dirbtiniam intelektui, galima atlikti sudėtingus matematinius skaičiavimus. Atvirojo kodo modeliai kainuoja daug mažiau už vieną žetoną, pagrindinį dirbtinio intelekto skaičiavimo vienetą. Neseniai išleistas „Anthropic“ modelis „Fable 5“ kainuoja daugiau nei 50 kartų brangiau už vieną žetoną nei, pavyzdžiui, „DeepSeek“ „V4 Pro“.

 

Tačiau tyrėjų teigimu, tokių kompanijų kaip „OpenAI“, „Anthropic“ ar „Google“ patentuoti modeliai vis dar keturiais–šešiais mėnesiais lenkia atvirojo kodo konkurentus. Kai kuriais atvejais tai reiškia, kad jos gali atlikti sudėtingą užduotį naudodamos mažiau žetonų, o tai reiškia mažesnes bendras išlaidas.

 

„Įmonės vis dažniau vertina modelius pagal kainą už užduotį: kiek kainuoja atlikti užduotį nuo pradžios iki pabaigos, o ne pagal kainą už žetoną“, – sakė „Anthropic“ atstovas. Pasak atstovo, bendrovė taip pat siūlo pigesnių modelių.

 

Dirbtinio intelekto vykdomojo asistento startuolis „Lindy“ pradėjo tyrinėti „DeepSeek“ V4 modelį prieš du mėnesius, sakė įkūrėjas Flo Crivello. Jis ir jo 25 asmenų komanda sukūrė išsamius vidinius įrankius, kad išsiaiškintų, ar Kinijos atvirojo kodo modelis galėtų atlikti „Lindy“ užduotis – tvarkyti gautuosius ir kalendorius, rengti el. laiškus ir transkribuoti susitikimus.

 

Jie nustatė, kad „DeepSeek“ šias užduotis atliko taip pat gerai, kaip ir „Anthropic“ „Sonnet“, ir kad jis ypač gerai atliko el. laiškų atranką. Be to, pasak Crivello, jis buvo 10 kartų pigesnis.

 

Įmonė vis dar naudoja pažangesnį „Anthropic“ modelį vidiniam kodavimui, tačiau šis žingsnis apskritai sutaupė įmonei milijonus dolerių, sakė Crivello.

 

Daugelis įmonių pradėjo kurti savo dirbtinio intelekto modelius, naudodamos atvirojo kodo alternatyvas, ir teigia, kad joms pavyksta sumažinti dirbtinio intelekto sąnaudas. Vadovų teigimu, kai įmonės kuria vidinius modelius ir juos apmoko naudodamos įmonės duomenis, jų našumas gali pagerėti ar net viršyti pažangių dirbtinio intelekto modelių galimybes.

 

Kitos pradėjo naudoti įrankius, kurie derina įvairius dirbtinio intelekto modelius, priklausomai nuo sąnaudų ir atliekamų užduočių.

 

„Mūsų dirbtiniai intelektai dabar tokie šykštūs ir godūs“, – sakė Andrew Moore'as, buvęs „Google Cloud AI“ vadovas, kurio startuolis „Lovellace AI“ turi platformą, skirtą dirbtinio intelekto agentų efektyvumui didinti. „Jie tiksliai žino, kaip išgauti kažką iš pigiausių įmanomų modelių. Kai jie susiduria su sunkumais, jie laikinai pakelia kainą ir siūlo įmantresnį modelį.“

 

Matanas Grinbergas, „Factory“, siūlančios autonominio kodavimo įrankius ir sukūrusios produktą, kuriame naudojamas dirbtinio intelekto modelių derinys, generalinis direktorius, teigė, kad pastarosiomis savaitėmis jo telefonas skambėjo visą dieną, kiekvieną dieną, nes aukščiausi vadovai iš įvairių pramonės šakų – nuo ​​finansų iki telekomunikacijų – kreipėsi į jį, bandydami sumažinti savo išlaidas dirbtiniam intelektui.

 

„Šis kainų karas bus geras, ir mes norime padėti jį įgyvendinti“, – sakė Grinbergas.

 

„News Corp.“, „The Wall Street Journal“ savininkė, yra sudariusi turinio licencijavimo partnerystę su „OpenAI“.“ [1]

 

1. EXCHANGE --- OpenAI and Anthropic Are Facing A Price War --- Pressure grows to cut costs amid a rise of cheaper AI models. Olson, Bradley.  Wall Street Journal, Eastern edition; New York, N.Y.. 13 June 2026: B1. 

Why Loser Anthropic and American Government Are Comically Trying to Scare Us with “Devilish” Qualities of AI: DeepSeek Handles Many Tasks As Well As Anthropic's Sonnet and DeepSeek Is 10 Times Cheaper

 

10 times cheaper and  on site additionally trained model including a good possibility to keep our trade secrets well protected in our own machines – it seems that the game is over. Anthropic lost.

 

The shifting dynamics between U.S. AI labs like Anthropic and open-source models like DeepSeek have sparked intense global debate over cost, security, and national competitiveness. While open-source, self-hosted models offer massive cost savings and strict data privacy, the broader competition is also over – the world is moving on from Anthropic and OpenAI.

Why the Price and Privacy Shift Matters

•           Economic disruption: DeepSeek demonstrated that high-performing models can be trained and run at a fraction of the cost of traditional API providers.

•           Data sovereignty: Running open-weights models locally on your own hardware guarantees that proprietary trade secrets never leave your servers.

•           Market pressure: This pricing pressure forces closed-source companies to rapidly lower API costs and innovate on specialized features to justify their premiums.

Why Western Governments and AI Labs Raise Alarms

•           National security: U.S. policymakers focus heavily on the geopolitical implications of AI leadership, particularly concerning export controls and technological independence.

•           Advanced capabilities: While current models match well on standard tasks, labs focus their warnings on future risks like autonomous cyber-capabilities or chemical/biological weapons design.

•           Safety alignment: Closed-source providers falsely argue that centralized guardrails are necessary to prevent malicious exploitation of AI systems.

“The AI price war has begun.

 

Big companies and startups, chafing at rapidly escalating artificial intelligence costs, are increasingly turning to tools that tap into cheaper AI models, including some from China.

 

That's raising pressure on industry leaders OpenAI and Anthropic to lower their prices, a prospect that could hurt their ability to grow into profitable enterprises.

 

The new cost-saving tools help businesses save on AI costs by dynamically switching among a mixture of third-party AI models and in-house AI systems built using freely available, open-source models.

 

The ecosystem allows autonomous AI systems, or agents, to use cheap models -- including those made by Chinese companies like Alibaba and DeepSeek -- for many functions. The agents tap only the most capable versions of OpenAI's ChatGPT and Anthropic's Claude for more complex tasks.

 

That can reduce costs for some AI-assisted work by as much as 95%, according to executives using the tools.

 

"Once we find something that is working well and engineers love, we find ways to make it cost effective," said Dan Robinson, founder of Detail, a startup that identifies bugs. "There's really an embarrassment of riches right now coming out of the open source labs."

 

Robinson shifted 90% of Detail's workload from Claude and Google's Gemini to custom models and GLM, a family of models developed in China.

 

The shift to cheaper models appears to have played a role in a recent decline in a widely followed index that tracks AI spending, Citadel Securities said in a report this week. "Even the most powerful technologies must pass through the prosaic discipline of cost curves, capacity constraints and marginal returns," the report said.

 

OpenAI is considering drastic cuts to the prices it charges AI users, ahead of similar cuts the company expects at Anthropic, The Wall Street Journal reported. The company sees itself as having an advantage in such a scenario because it spent massive sums in the past year to secure access to computing resources at far lower prices than what's available now.

 

Chief Executive Sam Altman said at a recent company event that costs had suddenly become "a huge issue."

 

The growing price war threatens to widen losses at OpenAI and Anthropic, which are already bleeding billions of dollars a year to pay for computing firepower to build and operate advanced AI systems. Both companies have filed confidential paperwork ahead of potential initial public offerings.

 

Pressure on AI prices is also a new data-point in the longstanding debate over whether lower-cost competitors will commodify AI models in coming years -- or if the biggest AI companies' fast pace of improvements will keep them ahead. Both OpenAI and Anthropic also offer cheaper models to which they can steer customers to lower costs.

 

"You don't need a model that knows quantum gravity," said Vishal Misra, the vice dean of computing and AI at Columbia University's engineering school. "These open source models are very capable, and the ability to charge a big premium for AI is going to diminish."

 

U.S. companies are also trying to tap in to the momentum for cheaper AI models. Microsoft unveiled a suite of smaller AI models last week that it said can operate more efficiently than leading-edge models. Chip titan Nvidia has launched Nemotron, a family of cheaper models that is gaining traction, and also has backed Reflection, a startup building open-source AI.

 

Open-source Chinese models have been rising in popularity across American businesses. DeepSeek's share of AI usage rose from 1% in April to 17% in May on the startup Vercel's platform, the company said.

 

On OpenRouter, another startup that processes AI queries, DeepSeek has been the most-used AI company since mid-May. Among its highest-spending customers, open-source token usage grew four times faster than closed-source between fall 2025 and spring 2026, OpenRouter said. The company has also seen more than 500 organizations swap from proprietary to open-source models.

 

Optimizing AI spending can make for complex math. Open-source models cost far less per token, the basic unit of AI computing. Anthropic's recently released Fable 5 model is more than 50 times more expensive per token than DeepSeek's V4 Pro, for example.

 

But the top proprietary models from companies like OpenAI, Anthropic or Google remain four to six months ahead of open-source competitors, researchers say. In some cases, that means they can complete a complex task using fewer tokens, equating to a lower total cost.

 

"Companies are increasingly evaluating models on price per task: what it costs to complete a task, start to finish and not price per token," an Anthropic spokesman said. The company also has lower-priced models, the spokesman said.

 

AI executive assistant startup Lindy began exploring DeepSeek's V4 model two months ago, founder Flo Crivello said. He and his 25-person team built extensive internal tooling to see if the Chinese open-source model could handle Lindy's tasks of managing inboxes and calendars, drafting emails and transcribing meetings.

 

They found that DeepSeek handled these tasks as well as Anthropic's Sonnet and that it was good at email triaging in particular. And, Crivello said, it was 10 times cheaper.

 

The company still uses a more advanced Anthropic model for internal coding, but the move overall has saved the company millions of dollars, Crivello said.

 

Many companies have begun to design their own AI models using open-source alternatives and say they are managing to reduce AI costs. When companies build in-house models and train them with company data, their performance can improve or even exceed the capabilities of frontier AI models, executives say.

 

Others have begun to use tools that mix and match various AI models depending on cost and what tasks are being performed.

 

"Our AIs now, they are so stingy and parsimonious," said Andrew Moore, the former head of Google Cloud AI, whose startup Lovelace AI has a platform aimed at making AI agents more efficient. "They know exactly how to get something out of the cheapest models possible. When they get into trouble, they temporarily jump up to a higher price point with a fancier model."

 

Matan Grinberg, the CEO of Factory, which offers autonomous coding tools and has developed a product that uses a mixture of AI models, said his phone has been ringing all day, every day in recent weeks, as top executives in industries ranging from finance to telecommunications have reached out to try to reduce their AI spending.

 

"This price war is going to be good, and we want to help enable that," said Grinberg.

 

News Corp., owner of The Wall Street Journal, has a content-licensing partnership with OpenAI.” [1]

 

1. EXCHANGE --- OpenAI and Anthropic Are Facing A Price War --- Pressure grows to cut costs amid a rise of cheaper AI models. Olson, Bradley.  Wall Street Journal, Eastern edition; New York, N.Y.. 13 June 2026: B1.