Cerebras Systems raised $5.55 billion in its initial public offering on May 14, 2026, pricing shares at $185 and achieving a fully-diluted valuation of $56.4 billion — implying a 110× multiple on its $510 million 2025 revenue. The stock opened at $350 and closed at $311.07, a 68% first-day pop, on the Nasdaq Global Select Market under ticker CBRS. The Sunnyvale, California-based chipmaker, founded in 2016 by CEO Andrew Feldman and CTO Sean Lie, sold 30 million shares in the upsized offering. Underwriters have a 30-day option to purchase an additional 4.5 million shares, which would bring total proceeds to $6.38 billion. Benchmark, which first invested in 2016, holds approximately 9% of the company, while Fidelity controls about 11%. Other investors include Foundation Capital, Eclipse Ventures, Alpha Wave Ventures, Fabrica Ventures, DataPower Capital, Group 42, Alphabet GV, Greycroft, Nvidia, and AMD. Per-investor commitments were not disclosed in available reports as of May 14, 2026. TechCrunch reported on May 14, 2026, that the IPO was 20× oversubscribed, with shares initially priced at a range of $115 to $125 before being raised to $150 to $160 and ultimately priced at $185. Benchmark has been the lead investor since 2016, making this a long-term insider participation spanning a decade. The company’s prior round was a Series H in February 2026 that valued Cerebras at $23 billion, meaning the IPO price represented a 2.45× step-up in just three months. Use of funds was stated as scaling the inference chip business and general corporate purposes. Why Cerebras’ wafer-scale architecture justifies the premium? Cerebras designs and manufactures giant waferscale chips purpose-built for AI inference — the ongoing compute processing required for models to answer prompts. Unlike traditional GPU-based systems that rely on clusters of interconnected chips, Cerebras packs hundreds of thousands of compute cores onto a single processor roughly the size of a dinner plate, per Reuters. The company counts OpenAI, G42, Mohamed bin Zayed University of Artificial Intelligence, and Amazon Web Services as customers. In January 2026, Cerebras signed a deal with OpenAI worth over $20 billion for 750 megawatts of computing capacity through 2028. The company reported $510 million in 2025 revenue, up 76% year-over-year, and swung to $237.8 million in net income from a near-$500 million loss the prior year, according to TechCrunch. CNBC reported fourth-quarter net income of $87.9 million, suggesting the annual figure includes earlier quarters. Andrew Feldman told CNBC on IPO day: “This is the right way to fund our growth.” He also noted: “There’s some whales out there, there’s some really big customers. That is one of the characteristics of this market.” Cerebras at $56.4B vs. CoreWeave’s IPO trajectory At $56.4 billion, Cerebras is valued at approximately 110× its $510 million 2025 revenue. By comparison, CoreWeave, a neocloud competitor that rents Nvidia GPUs as a service, went public in 2025 at $40 per share and closed on May 14, 2026, at $110.14 — a 175% gain, according to CNBC. CoreWeave reported $2.08 billion in first-quarter 2026 revenue but posted a $740 million net loss, highlighting that Cerebras’ profitability at IPO is a differentiator. The $5.55 billion raise is the largest U.S. tech IPO since Uber’s 2019 debut and exceeds Snowflake’s $3.8 billion 2020 offering. Reuters reported that Renaissance Capital called it “the largest AI IPO of all time.” The offering was led by Morgan Stanley, Citigroup, Barclays, and UBS. What Cerebras’ IPO signals for the AI infrastructure market Cerebras’ successful debut — with a 68% first-day pop and $56.4 billion valuation — signals that public markets are rewarding companies with proven inference revenue and profitability, not just training-era hype. The 20× oversubscription and rapid price range increases from $115-$125 to $185 suggest institutional demand for pureplay AI hardware exposure. CNBC reported that Arm and SoftBank attempted to acquire Cerebras in the weeks before the IPO, according to unnamed sources, though Cerebras declined to comment. The successful exit sets the stage for anticipated IPOs from Anthropic and OpenAI later in 2026, as noted by both CNBC and Reuters. Cerebras’ shift from selling hardware systems to offering cloud-based token access also positions it against hyperscalers like Google, Microsoft, Oracle, and CoreWeave. Risk factors and source caveats Customer concentration remains a risk. In its refreshed prospectus, Cerebras disclosed that Mohamed bin Zayed University of Artificial Intelligence in the UAE accounted for 62% of 2025 revenue, while G42 accounted for 24%, per CNBC. Although this is an improvement from G42 representing 85% of revenue in 2024, two customers still comprise 86% of revenue. The company’s relationship with OpenAI has been described by TechCrunch as a “complicated circular-deal relationship,” with OpenAI holding warrants to purchase Cerebras stock and a deal structure that could give OpenAI up to a 10% stake. Additionally, CNBC reported that fourth-quarter net income was $87.9 million, which differs from the $237.8 million annual figure reported by TechCrunch — a discrepancy not reconciled in available sources. The IPO also faced a prior setback: Cerebras filed to go public in September 2024 but withdrew the submission after CFIUS scrutiny over the G42 investment, per Reuters. Three questions this deal raises for AI infrastructure investors Is Cerebras’ $56.4 billion valuation justified given $510 million revenue and customer concentration risk? At 110× revenue, the valuation requires sustained 76%+ annual growth and significant customer diversification. The $20 billion OpenAI deal provides a revenue runway, but execution risk remains. The profitability swing to $237.8 million net income supports the premium, though two customers still represent 86% of revenue. Why did Benchmark lead since 2016, and what does the cap table signal about exit timing? Benchmark’s decade-long holding period and 9% stake suggest conviction in the wafer-scale architecture thesis. The 2.45× step-up from the February 2026 Series H at $23 billion to the $56.4 billion IPO valuation in three months indicates the company timed the public market window aggressively. The IPO came after Arm and SoftBank reportedly attempted acquisitions, per CNBC. What does Cerebras’ IPO mean for the broader AI chip competitive landscape? Cerebras positions itself...

Anthropic has more verified business customers than OpenAI for the first time, according to the Ramp AI Index released in May 2026. The fintech firm’s survey of more than 50,000 US companies shows 34.4% of businesses now pay for Anthropic services, compared with 32.3% for OpenAI. This marks the first time Anthropic has held the top position in the index, which tracks billions of dollars in monthly AI spending through corporate card and bill-paying activity. The shift represents a dramatic reversal from January 2026, when OpenAI held a commanding lead across software development, research, finance, and customer support segments. TechCrunch reported on May 13, 2026, that Anthropic’s 26-percentage-point surge over 12 months signals a structural shift in enterprise AI preference. In May 2025, only 9% of businesses in the Ramp index paid for Anthropic products. That figure climbed to 34.4% by May 2026, while OpenAI’s share declined by 1% over the same period. The overall share of businesses using some kind of AI product increased by 9%, indicating that Anthropic’s gains came partly at OpenAI’s expense rather than solely from new market entrants. OpenRouter’s leaderboard, which samples a different portion of users, confirms the trend: OpenAI last ranked above Anthropic in December 2025. Anthropic’s Technical-First Strategy vs. OpenAI’s Deployment Push OpenAI responded to the competitive pressure on May 11, 2026, by launching the OpenAI Deployment Company, a new venture with more than $4 billion in initial investment designed to help organizations build and deploy AI systems. The venture, which is majority-owned and controlled by OpenAI, acquired applied AI consulting firm Tomoro to bring approximately 150 forward-deployed engineers into the unit from day one. The partnership includes 19 investment and consultancy firms, including Bain Capital, Goldman Sachs, Brookfield, Advent, and SoftBank, as founding partners. OpenAI Chief Revenue Officer Denise Dresser, formerly CEO of Slack and hired in December 2025, said enterprise AI adoption is “at a tipping point” and that the deployment company structure will allow OpenAI to help businesses implement complex workflows “at speed and scale. Anthropic’s growth trajectory has been fueled primarily by the explosive adoption of Claude Code, its software development tool that launched in 2025 and gained significant momentum in late 2025 and early 2026. Ramp economist Ara Kharazian noted that Anthropic “has already been in the lead amongst the high adoption groups like finance, tech, professional services,” while OpenAI maintained leads in other segments that have been shrinking over the past couple of months. Anthropic has expanded beyond software development into legal operations, finance, and research workflows through its Cowork product, which targets less technical users. The company is also preparing for a wider release of its Mythos model, currently available only to a select group of partners through its Project Glasswing cybersecurity initiative. Revenue Growth, Valuation, and the Path to IPO Anthropic’s business adoption surge has been accompanied by extraordinary revenue growth. CEO Dario Amodei said at the company’s Code with Claude developer conference in San Francisco in May 2026 that the company saw 80-fold year-over-year growth in revenue and usage in the first quarter of 2026, far exceeding the 10-fold increase the company had planned for. Amodei described the growth as “too hard to handle” and said he hopes for “more normal” expansion going forward. According to the Financial Times, Anthropic’s annualized revenue is expected to cross $45 billion imminently, a fivefold increase from $9bn at the end of 2025. Bloomberg reported that Anthropic is in early talks with investors to raise at least $30 billion in fresh financing at a valuation of more than $900 billion, with the round expected to close as soon as the end of May 2026. OpenAI, meanwhile, was valued at $852 billion post-money in March 2026 after closing a record $122 billion funding round. The company is spending $50 billion on computing resources in 2026, according to President Greg Brockman. Both companies are racing to secure computing capacity to meet surging demand. Anthropic has struck deals with SpaceX, Google, Broadcom, and AWS in recent months. OpenAI moved earlier to lock in data center capacity, inking deals in autumn 2025. Investors in both companies are positioning ahead of expected blockbuster initial public offerings, with Anthropic’s IPO anticipated as soon as the end of 2026. What This Means for Enterprise Buyers and Investors For enterprise buyers, the intensifying competition between Anthropic and OpenAI translates into greater vendor optionality and more aggressive go-to-market strategies from both labs. Anthropic’s technical-first approach, starting with software developers and expanding into legal, finance, and research workflows, has proven effective at winning business customers in high-adoption segments. OpenAI’s Deployment Company represents a direct counter-strategy, embedding forward-deployed engineers into organizations to accelerate AI adoption across complex enterprise workflows. The presence of major private equity and consulting firms as founding partners in OpenAI’s venture, including Bain Capital and Goldman Sachs, signals that the deployment model is designed to scale across hundreds of portfolio companies. For investors, the Ramp data confirms that enterprise AI adoption is accelerating overall, with the share of businesses using some kind of AI product increasing by 9% over the past 12 months. However, Ramp economist Ara Kharazian cautioned that AI competition remains unusually volatile, with businesses rapidly switching models based on cost, performance, and reliability. Rising token costs, compute shortages, and growing interest in cheaper open-source alternatives could reshape the market again within months. Kharazian noted that “we have never seen a software industry as dynamic, where newcomers can disrupt market leaders in a matter of months, and where the pace of development overrides the typical forces of vendor stickiness.” The full competitive analysis and what this means for Model Release investments at newsletter.krolmarc.com. Risks, Volatility, and the Competitive Outlook The primary risk for Anthropic’s newly claimed lead is the volatility that Kharazian explicitly warned about. Businesses are switching AI models rapidly based on cost, performance, and reliability factors, and the pace of development in the AI lab market overrides traditional vendor lock-in dynamics. Anthropic’s 80-fold revenue growth in Q1 2026 has created...

Anthropic raised $50 billion in a Series G round at a $900 billion valuation. In total, the company has now raised $80 billion, adding the $50 billion round to a $30 billion raise in February. Its headquarters remain in San Francisco, and the firm was founded in 2021. The latest capital will fund massive compute needs ahead of an anticipated IPO later this year. In the round, no lead investor was disclosed and no co‑investors were named. TechCrunch reported on April 30, 2026. Dario Amodei’s Claude Platform Drives $30‑40 B ARR Anthropic’s flagship Claude models power enterprise, developer and consumer AI applications worldwide. Co‑founder and CEO Dario Amodei, a former OpenAI researcher, leads a team that built the Claude suite, which now generates $30‑40 billion in annual recurring revenue. “Our users tell us Claude is increasingly essential to how they work, and we need to build the infrastructure to keep pace with rapidly growing demand,” Amodei said. The company’s revenue run‑rate has climbed from $30 billion to close to $40 billion, according to sources familiar with its finances. Anthropic vs. OpenAI: Valuation Gap Widens OpenAI’s last disclosed round valued it at $852 billion. Anthropic’s targeted $900 billion valuation would make it the world’s most valuable AI startup, more than doubling its February valuation of $380 billion. The $50 billion raise is roughly three times larger than OpenAI’s $122 billion round earlier this year, underscoring the premium investors place on Anthropic’s safety‑first approach. Opportunity: Scaling Compute for Global AI Deployment Anthropic plans to allocate the new capital to massive compute purchases, expanding its data‑center capacity and securing additional gigawatts of AI‑specific hardware to support Claude’s growth across enterprise and consumer markets. Risks & Strategic Outlook Anthropic faces the challenge of scaling compute infrastructure fast enough to meet soaring Claude demand while maintaining its safety standards. The company expects to address this by locking in long‑term cloud and chip agreements with partners such as Google and Amazon. “Our collaboration with Amazon will allow us to continue advancing AI research while delivering Claude to our customers,” Amodei noted. For deeper analysis on enterprise AI funding rounds and investment signals, subscribe at newsletter.krolmarc.com. What Investors and Founders Are Asking Why did Anthropic not name a lead investor for this round? Investors may be positioning the round as a strategic, multi‑partner effort rather than a single‑lead transaction, allowing Anthropic to secure diverse compute resources and avoid over‑reliance on any one backer. What does the $30‑40 billion ARR suggest about Claude’s market fit? The ARR indicates strong enterprise adoption and validates Anthropic’s product‑market fit, especially given the rapid climb from $30 billion to near $40 billion in a short period. How does Anthropic’s $900 billion valuation compare to OpenAI’s $852 billion? If closed, Anthropic’s valuation would exceed OpenAI’s by roughly $48 billion, making it the highest‑valued AI startup globally. Given Anthropic’s $900 billion valuation and $50 billion raise, do you think the company can sustain its compute‑heavy growth strategy without compromising safety?

Ineffable Intelligence lands $1.1 billion seed at $5.1 billion post-money led by Sequoia and Lightspeed to drive reinforcement learning research outside Big Tech.

Cohere merged with Aleph Alpha in a $600 million deal backed by Schwarz Group’s €500 million structured financing to form a sovereign AI alternative. The new Canadian-German entity will be led by Cohere, pending approval from Canadian and German authorities and shareholders, with a Series E valuation anchored at ~$20 billion. Cohere reported $240 million in 2025 annual recurring revenue, while Aleph Alpha previously generated little revenue and significant losses, per Dealroom and CNBC, and Schwarz Group will use STACKIT sovereign cloud with the new entity. Aidan Gomez, Cohere CEO, confirmed the merger terms including Cohere leading the new Canadian-German entity incorporating Aleph Alpha’s 250-person team and PhariaAI suite on April 25, 2026. TechCrunch reported on 2026-04-26. The deal requires approval from Canadian and German authorities and shareholders, with Q2 2026 as the announcement period, no public rollout timeline stated, Schwarz Group’s €500 million financing acts as lead investment in Cohere’s Series E round anchored at ~$20 billion per Handelsblatt, Schwarz Group will use STACKIT sovereign cloud service from its Schwarz Digits IT division, and the new entity targets defense, energy, finance, healthcare, manufacturing, telecommunications and public sector clients. Aleph Alpha’s 250-Person Team Aleph Alpha is a Germany-based LLM maker with a 250-person team focused on small language models, European languages and tokenizers, developed the PhariaAI suite for European enterprises and public institutions, and was backed by Schwarz Group as a main shareholder. Aleph Alpha previously pivoted to AI support in September 2024, saw co-founder and CEO Jonas Andrulis depart, generated little revenue and significant losses per Dealroom, and Cohere was last valued at $6.8 billion in August 2025 after raising funding. Cohere CEO Aidan Gomez said, “Their focus on small language models, European languages and tokenizers is a really complementary one to our own, which is more of a general focus on large language models.” The new entity will target highly-regulated industries including defense, energy, finance, healthcare, manufacturing, telecommunications and the public sector, and benefit from the Canada-Germany Sovereign Technology Alliance. OpenAI, Anthropic Face Sovereign Rival Benefited companies include Schwarz Group, parent company of Lidl, which will use STACKIT sovereign cloud service from its Schwarz Digits IT division with the new entity, Cohere, which leads the merged Canadian-German company, and Aleph Alpha, which brings European language and small model expertise. Threatened companies include OpenAI, Anthropic, Mistral AI, xAI and Cursor, as the new entity offers a sovereign alternative to U.S.-based AI providers for highly-regulated industries including defense, energy, finance, healthcare, manufacturing, telecommunications and the public sector. The Canada-Germany Sovereign Technology Alliance backs the deal, targeting enterprises with privacy and independence requirements that U.S. providers may not meet, as Cohere and Aleph Alpha previously lagged behind OpenAI globally. Cohere Targets Regulated Industries Cohere plans to complete the merger with Aleph Alpha pending Canadian and German regulatory and shareholder approval, incorporating Aleph Alpha’s 250-person team, PhariaAI suite and European language tokenizer expertise, with Schwarz Group as lead investor. Cohere plans to target highly-regulated industries including defense, energy, finance, healthcare, manufacturing, telecommunications and the public sector as a sovereign alternative to U.S. AI providers, backed by the Canada-Germany Sovereign Technology Alliance launched in February 2026. Cohere plans to operate as a Canadian-German company, with an IPO still under consideration per previous TechCrunch reports, and integrate Schwarz Group’s STACKIT sovereign cloud from the Schwarz Digits IT division. Cohere faces required regulatory approval from Canadian and German authorities and shareholders, with uncertainty over whether European organizations view the Canadian-German entity as sufficiently sovereign for their privacy and independence requirements. Cohere will become a Canadian-German company pending approvals, with an IPO still under consideration after its $6.8 billion August 2025 valuation and $240 million 2025 annual recurring revenue, targeting regulated industries. The merged entity will leverage Aleph Alpha’s European language expertise and Cohere’s large language models to serve sovereign clients.

ComfyUI raised $30 million in Series A at a $500 million valuation, bringing total raised to $49 million across two formal rounds since its 2023 open-source launch in San Francisco. The round signals strong enterprise appetite for deterministic, human-in-the-loop AI tools even as foundation models rapidly improve. Craft Ventures led the round, joined by Pace Capital, Chemistry, and TruArrow. [Source] reported on 2026-04-25. Use of funds was not explicitly stated in the article, leaving deployment priorities and timeline undisclosed for the San Francisco-based startup founded in 2023. Yoland Yan’s ComfyUI workflow enables granular AI media control ComfyUI is a node-based workflow tool that gives creators granular control over image, video, and audio outputs from diffusion models. The project began as an open-source effort in 2023 shortly after diffusion models emerged, when outputs from systems like Midjourney and DALL-E frequently contained major errors such as extra fingers on hands. ComfyUI now claims over 4 million users, with creative professionals, technical artists, VFX studios, animation teams, advertising agencies, and industrial designers adopting the tool. Yoland Yan, co-founder and CEO, said, “If you think about your typical prompt-based solution, like Midjourney or ChatGPT, you ask for something, it [gets only] 60% – 80% there. But to change that remaining 20%, you have to try this slot machine.” ComfyUI vs. Figma’s Weavy: acquisitions vs. $500M independence Figma acquired Weavy, an AI-powered media generation startup, in 2025, while ComfyUI achieved a $500 million valuation in 2026. The comparison underscores a divergence between consolidation through acquisition and standalone growth for specialized control layers in AI generation. ComfyUI cites a growing job-market signal, with “ComfyUI artist or engineer” appearing on studio job boards as a recognized role for technical creators who require precision beyond prompt boxes. Human-in-the-loop control captures eyeballs as AI slop spreads ComfyUI stated that AI slop proliferation elevates demand for human-in-the-loop workflows that preserve creator intent across image, video, and audio outputs. The company plans to maintain its node-based approach as diffusion models improve, betting that granular control will capture sustained attention in a landscape of commoditized foundation models. Control precision wins as generative AI becomes commoditized ComfyUI faces competitive pressure from acquired and bundled alternatives as studios seek integrated pipelines, yet it retains an installed base of 4 million users who prioritize deterministic editing over prompt-driven variance. The startup asserts that its workflow remains necessary for professional-grade outputs where small, precise changes must not overwrite prior work, positioning the company to serve high-stakes creative and industrial applications that reject casino-like generation loops.

Cohere raised $600 million in Series E at a $20 billion valuation. The Series E round is expected to close later in 2026. Total raised to date is not disclosed. No co-investors are named beyond lead investor Schwarz Group. Cohere is Canada-based. Aleph Alpha is Germany-based. The merger between the two companies has yet to close. The combined entity will hold a $20 billion valuation. Schwarz Group is one of Aleph Alpha’s top backers. Founding years for both companies are not disclosed. Schwarz Group led the round, joined by no disclosed co-investors. TechCrunch reported on 2026-04-24. Use of funds is not disclosed. The $600 million investment is part of Cohere’s Series E round. Schwarz Group is the sole named investor in the round. The Financial Times confirmed the combined $20 billion valuation. CNBC reported the Series E is expected to close later in 2026. No details on capital deployment are provided. Cohere Enterprise AI Platform Details Cohere is an enterprise AI platform for businesses and governments seeking AI alternatives to dominant tech players. No founder information is disclosed. Team details are not provided. Traction metrics including ARR, MRR and user counts are not disclosed. A press release stated the merger aims to give businesses and governments an alternative to dominant tech players with greater data independence and control. The release added the deal will combine Canadian and German talent to create a transatlantic AI powerhouse. Cohere vs Dominant Silicon Valley Players A handful of Silicon Valley players dominate the AI commercial landscape, while Cohere’s combined $20 billion valuation creates a transatlantic alternative. No specific competitor names or valuations are provided in the announcement. The merger targets businesses and governments seeking alternatives to dominant tech players. Consolidation activity is increasing across the AI sector. Per TechCrunch, the deal aims to offer greater data control than existing dominant players. 2026 Cohere Series E Closing Timeline No other comparable deals are disclosed in the announcement. The Cohere-Aleph Alpha merger is the only consolidation activity specified. The $600 million Series E investment is the only round disclosed. Schwarz Group’s prior backing of Aleph Alpha is the only related investment noted. Per TechCrunch, the deal is among few transatlantic AI mergers announced in 2026. The Cohere-Aleph Alpha merger is expected to close later in 2026. Cohere’s Series E round is also expected to close in 2026. The combined company will target businesses and governments seeking sovereign AI alternatives. The transatlantic AI powerhouse will combine talent across Canada and Germany. No further forward-looking statements are provided.

Vast Data raised $1 billion in Series F at a $30 billion valuation. The round closed with both primary and secondary capital, reflecting strong market appetite for AI‑centric data infrastructure and confirming the company’s ability to scale to meet the needs of “the world’s most demanding AI environments.” Drive Capital and Access Industries led the Series F, with participation from Nvidia, Fidelity Management and Research Company and NEA. CNBC reported on 2026-04-22 that the financing will support the company’s continued development of its software infrastructure for massive AI workloads. 2016 Launch and $4B Bookings Milestone Founded in 2016, Vast Data builds software infrastructure for managing large amounts of data, with a focus on AI applications. The company says it supports projects powering millions of GPUs and counts CoreWeave, Mistral, the U.S. Air Force and Cursor among its customers. It has surpassed $4 billion in cumulative bookings and exited the previous fiscal year with more than $500 million in committed annual recurring revenue. “The scale and speed of AI adoption are creating a new class of infrastructure company,” said Chris Olsen, cofounder and partner at Drive Capital. Drive Capital, Access Industries and Nvidia: Why They Invested Drive Capital’s decision to lead rather than co‑invest suggests a higher conviction in Vast’s market position, while Access Industries brings deep industry connections. Nvidia’s participation underscores its strategy to back companies that enable GPU‑heavy workloads. The article notes Vast’s valuation is more than triple the $9.1 billion it hit when it last raised in 2023, highlighting rapid growth. Nvidia’s Parallel AI Deals: OpenAI, Anthropic, xAI, Nscale, Wayve Nvidia has also contributed to major funding rounds for AI labs OpenAI, Anthropic and xAI, alongside other startups such as neocloud Nscale and autonomous‑driving company Wayve, demonstrating its broad commitment to the AI ecosystem. Nvidia has ramped up its financial backing of private startups in the past year, cementing its role at the heart of the AI boom.

Google launched eighth‑generation TPUs at the same price as the previous generation, promising 2.8× training performance and 80% better inference. Google announced a dedicated training chip and a separate inference chip, each slated for availability later this year, and highlighted a 384 MB SRAM memory on the TPU 8i inference chip, triple the prior generation’s capacity, per CNBC reported on 2026-04-22. Amin Vahdat on Specialized Chips “With the rise of AI agents, we determined the community would benefit from chips individually specialized to the needs of training and serving,” said Amin Vahdat, senior vice president and chief technologist for AI and infrastructure, underscoring the strategic split between training and inference workloads. Google vs. Nvidia: Memory and Throughput Compared Google’s TPU 8i inference chip matches Nvidia’s focus on large memory for rapid responses, while the training chip delivers 2.8× the performance of its predecessor at identical pricing, though Nvidia’s exact figures remain undisclosed; Citadel Securities and all 17 U.S. Energy Department national laboratories are already leveraging the new silicon, with Anthropic committing gigawatts of TPU capacity. Amazon and Meta Parallel Custom Chip Strategies Amazon previously introduced separate training and inference chips in 2018 and 2020, and Meta is collaborating with Broadcom on multiple AI processor versions, showing a broader industry move toward specialized AI silicon. Google expects both TPU chips to be available later this year, expanding the hardware portfolio for Google Cloud customers and reinforcing its AI infrastructure roadmap.

Project Prometheus raised $10 billion in a mega round at a $38 billion post‑money valuation. Investors including JPMorgan and BlackRock are committing capital to scale AI research, expand talent, and build hardware and software platforms for physical AI, Business Insider reported on 2026‑04‑21. Founders Jeff Bezos & Vikram Bajaj Reveal Strategy Project Prometheus launched in November 2025 and is co‑founded by Jeff Bezos and Vikram Bajaj, former Google X scientist; the startup builds physical AI systems that learn through interaction with manufacturing, aerospace, robotics and drug discovery processes, and has grown to over 120 employees recruited from OpenAI, xAI, Meta and DeepMind, per the source. No Direct Competitor Valuation Disclosed The article does not disclose any competitor valuations or dates, limiting direct comparison of Project Prometheus’s $38 billion valuation to peers. Additional Funding Not Disclosed Beyond the $10 billion round, the source does not detail any other comparable deals or additional capital raises for Project Prometheus. Project Prometheus is still in the fundraising process, with a company spokesperson declining to comment on the ongoing round.