Meta's Iris AI chip enters production in September 2026, aiming to cut NVIDIA reliance

Meta has announced that its custom AI chip, named Iris, will enter production in September 2026. The chip is part of Meta's MTIA (Meta Training and Inference Accelerator) program, developed in partnership with Broadcom and TSMC. Iris is designed to handle inference workloads and potentially lighter training tasks, aiming to reduce Meta's dependence on NVIDIA GPUs for these functions.

Specs, testing, and iteration

Meta says Iris follows a 6-week testing cycle and a 6-month iteration cadence, which the company claims allows it to adapt quickly to evolving AI model requirements. However, significant specifications such as performance benchmarks, power consumption, and interconnect bandwidth have not been disclosed. Without these numbers, Iris cannot yet be compared directly to existing products like NVIDIA's H100 or Google's TPU v5p.

Production timeline and context

Production begins in September 2026, aligning with Meta's target of reaching 7 GW of compute capacity by 2026. The company's capital expenditure plan is $125-145 billion, primarily allocated to compute infrastructure. Iris manufacturing will use TSMC's 5nm or 3nm nodes.

Competitive landscape

Meta is entering a crowded field of hyperscaler custom silicon. Google's TPU line, now on its fifth generation, has a multi-year lead in software maturity and ecosystem tools. Amazon's Trainium and Inferentia chips serve both internal AWS workloads and external customers. Microsoft's Maia 100, designed for Azure, refreshes on a 12-18 month cycle—slower than Meta's claimed 6-month cadence.

NVIDIA still dominates training workloads, and Meta continues to use H100s extensively for training its Llama models. Meta also uses AMD's MI300X in some clusters. Startups like Groq and Cerebras focus on low-latency inference but lack Meta's vertical integration advantages.

Software and reliability challenges

The software stack remains the critical question. NVIDIA's CUDA ecosystem is a deep moat, and Meta's custom stack must integrate smoothly with PyTorch to be useful. The company's first-generation MTIA chip, announced in 2023 on 7nm, had limited performance and software maturity. Meta's track record with custom silicon is mixed—its Reality Labs division has faced hardware productization challenges, though the MTIA team is separate and more focused.

A 6-month iteration cadence could lead to instability in datacenter deployments, such as compatibility breaks or thermal and power issues. Google's longer TPU cycle gives time for validation and ecosystem maturation. Without benchmarks and real-world testing, Iris remains a strategic hedge rather than a confirmed competitor.

Analysis

Meta's aggressive 6-week testing cycle and 6-month iteration cadence are ambitious for datacenter hardware, where reliability and backward compatibility are critical. Rapid iteration can introduce risks: half-baked revisions could undermine the vertical integration advantage if software integration lags behind. The company's $125-145 billion capex plan suggests it can afford to iterate, but without disclosed specifications or benchmarks, Iris is not yet a viable alternative to NVIDIA for heavy training workloads. The real test will be whether Meta can deliver a mature software stack and stable silicon at scale, not just faster hardware.

Read More

Meta’s ‘Iris’ AI Chip Enters Production in 2026, Signaling Push to Reduce NVIDIA Dependence

San Francisco, CA — Meta says its homegrown AI chip, code-named Iris, will enter production in September 2026. The chip is part of Meta’s broader push to design its own silicon for AI workloads, reducing reliance on external vendors like NVIDIA.

The company confirmed the timeline as part of a presentation to employees earlier this week. Iris is the latest in Meta’s MTIA (Meta Training and Inference Accelerator) program, which began in 2023. It is designed in partnership with Broadcom and will be manufactured by Taiwan Semiconductor Manufacturing Company (TSMC).

What Iris Is Designed to Do

According to Meta, Iris is built to handle three main tasks: training large AI models, running recommendation and ranking systems, and performing inference—the process of using a trained model to generate outputs. Meta says one Iris chip passed testing in just six weeks, a relatively fast turnaround for a custom semiconductor. The company plans to release a new chip roughly every six months through 2027.

Unlike NVIDIA’s H100 or B200, which serve as general-purpose accelerators for a wide range of workloads, Iris is purpose-built for Meta’s specific needs. That includes the company’s massive recommendation systems—the engine behind content ranking across Facebook, Instagram, and other Meta properties.

Meta estimates it will operate 7 gigawatts of compute capacity in 2026, double that in 2027. The company’s 2026 capital expenditure is projected between $125 billion and $145 billion, much of which will go toward AI infrastructure. That includes its own chips as well as deals with AMD for Instinct GPUs and Amazon for its homegrown CPUs.

Market Context: Custom Silicon Gains Traction

The AI chip market is projected to be worth between $130 billion and $160 billion by 2026. While NVIDIA is still expected to hold 65–70 percent of that market, the rise of custom silicon from cloud hyperscalers is reshaping the landscape. Google has its TPU line, Amazon offers Trainium chips, and Microsoft recently unveiled its Maia series.

Custom chips are projected to account for 15–20 percent of AI chip shipments by unit volume in 2026. That’s still a small share, but analysts say it’s meaningful for companies running at hyperscale. For Meta, the math is simple: designing a chip tailored to its recommendation inference workloads could offer better performance per watt and per dollar than a general-purpose GPU.

Inference is expected to be 60 to 70 percent of total AI compute demand in 2026, according to industry forecasts. That shift from training to inference is a key reason hyperscalers are investing in custom silicon. Training requires maximum throughput, but inference demands low latency and high efficiency—areas where a purpose-built chip can excel.

Meta has a specific advantage here: its recommendation systems are among the largest in the world, processing billions of ranking requests daily. A chip designed to optimize those specific operations could yield meaningful cost savings. But Meta also faces a disadvantage: unlike Google or Amazon, it does not sell cloud compute services. So there is no external revenue stream to offset the billions spent developing its own chips.

Supply Chain and Geopolitical Risks

TSMC’s advanced packaging technology, CoWoS, is a critical component for AI chips, and it remains a supply bottleneck. Meta declined to say if Iris would require CoWoS packaging or what capacity it has secured. The company also faces geopolitical risk: TSMC’s factories are concentrated in Taiwan, which is subject to potential disruption from China. Meta has not disclosed any contingency planning regarding alternative fabrication or packaging sources.

The rapid iteration cycle—a new chip every six months—suggests Meta is prioritizing time-to-market over perfection. That’s a different approach than NVIDIA’s, which typically refreshes architectures every one to two years. It also signals that Meta views AI silicon as an area of strategic urgency, not just operational efficiency.

Competition on Multiple Fronts

Meta’s Iris chip will not directly compete with NVIDIA in the broader market. But it does mean NVIDIA loses a high-volume customer for certain workloads. Meta still uses NVIDIA GPUs for large-scale training, and its AMD and Amazon deals provide additional flexibility. The company is essentially hedging against any single vendor’s pricing, availability, or performance limitations.

For Broadcom, the partnership is a major win. The company has been expanding its custom chip business, and helping Meta design a high-volume AI chip strengthens its position. For TSMC, every new custom chip from a hyperscaler adds to its already stretched capacity, but it also locks in long-term demand.

Analysis

Meta’s Iris chip is a bet on vertical integration—and a recognition that AI compute costs are rising faster than revenue growth. The company’s decision to iterate every six months, rather than annual or biennial cycles, tells you something important: Meta wants options. It wants to be able to redirect its own compute capacity without paying NVIDIA’s margins, and it wants to be fast enough to adapt as AI models evolve.

But there are real risks. Designing a chip is hard. Designing one at hyperscale—and doing so every six months—is exponentially harder. Meta has little public track record with silicon, and one successful test chip does not make a reliable product line. The six-month cadence could lead to rushed designs or quality issues. And if TSMC’s capacity constraints worsen, even a great chip is just a paperweight.

The bigger question is whether Meta’s chip will actually save money. Custom silicon only pays off at high volumes, and Meta has the volumes. But inference workloads evolve quickly as models change. A chip designed for today’s recommendation engine might not be optimal for tomorrow’s multimodal model. Meta is betting it can move fast enough to keep pace. That’s a bold claim—and one that only time, and billions in capex, will verify.

Read More

Findability Sciences Launches Rapid AI Readiness Tool for Dairy Plants

Findability Sciences, a SoftBank Group-backed AI company, has launched a self-serve diagnostic tool specifically for dairy processing plants. The LactaAI Discovery and Readiness Assessment is designed to help plant managers quickly identify where value is being lost and whether their existing systems are ready for AI integration.

Announced on May 19, 2026, in India, the tool aims to address what Findability calls the "data-to-decision gap" — the disconnect between operational technology (like PLCs and SCADA) and IT systems (like ERP and LIMS). Industry reports from USDA and IDF suggest that 40-50% of food processors lack this integration, making AI adoption challenging.

What It Does

The assessment provides three specific outputs:

• Identifies areas of value leakage across yield, energy, downtime, quality, and reporting
• Determines if existing plant systems — PLCs, SCADA, MES, ERP, LIMS — are "AI-ready"
• Recommends a starting point for fastest return on AI investment

According to Findability, the entire process takes minutes, not the weeks or months typically required for consultant-led assessments.

Reference Results

The company cites specific performance gains from prior deployments:

• Yield improvement: 0.4–0.6%
• Energy recovery in utilities: 8–15%
• Time-to-value: 6–10 weeks
• Estimated annual value for large dairy operations: USD 1 million to USD 4 million per plant

These figures are not extraordinary by industry standards — similar improvements are achievable with basic AI optimization — but they represent a tangible, verifiable baseline rather than a hyperbolic claim.

Platform Details

The LactaAI platform covers milk, cheese, whey protein, lactose, drying, packaging, utilities, quality, and enterprise operations. It is structured in two layers: Lacta Insight (plant floor) and Lacta BPC (business layer). Findability Sciences also offers a broader platform including forecasting tools, business co-pilots, and an agentic workflow engine built on its I-CUPP framework.

How It Compares

No direct competitor offers a rapid self-serve diagnostic tool specifically for dairy AI readiness. However, several alternatives occupy adjacent space:

• Rockwell Automation (Plex, FactoryTalk) — dominant in plant-floor integration, but their smart manufacturing assessments are consultant-led and take weeks.
• ABB (Ability Genix) — offers process optimization and condition monitoring, but readiness assessments are custom and include hardware audits.
• Tetra Pak (PlantSecure) — has deep dairy domain expertise but their AI play is via partners or internal analytics, not a self-serve product.
• McKinsey, BCG, Deloitte — paid, lengthy digital maturity assessments priced at tens to hundreds of thousands of dollars.

Findability's differentiator is speed and domain specificity. The tool explicitly targets dairy plant managers who are risk-averse to lengthy vendor engagements. For operators hesitant about AI due to long discovery cycles, this tool lowers the bar to a decision.

Company Background

Findability Sciences was founded in 2011, is headquartered in Burlington, Massachusetts, and has offices in Mumbai and Chhatrapati Sambhajinagar. The company serves over 50 enterprise clients across more than 250 deployments. It was named to Fortune's America's Most Innovative Companies list in 2023 and 2024 — a PR metric that indicates sustained media attention, though not necessarily market success.

What's Unclear

Several important details remain unknown:

• Pricing of the assessment — whether it's free, a one-time fee, or a subscription; this makes it hard for plant managers to budget for it
• Availability outside India — the initial launch appears focused on India; no mention of US or European markets
• Specific client testimonials or named reference plants — the claims are currently unverified
• Technical methodology — it's unclear whether the assessment is a questionnaire-based rules engine, an actual data screening from PLC/SCADA, or a simulated model; the "minutes" timeframe suggests a questionnaire approach, which may miss nuanced system integration

Analysis

The LactaAI Discovery and Readiness Assessment is a credible, productized lower-friction entry point for dairy AI adoption. The reference data is realistic — 0.4-0.6% yield improvement and 8-15% energy recovery are typical for baseline AI in dairy processing, not groundbreaking. That's actually reassuring: it suggests the company isn't overselling.

The biggest risk is adoption. Dairy plant managers are notoriously skeptical of software vendors. The tool may get a "quick look," but conversion to paid deployments will require proof points — ideally named plants with measurable ROI. Without a clear pricing model or at least one verifiable reference, the assessment risks being a free lead-gen tool with no next step.

Another risk: competitive response. Rockwell, Siemens, or Tetra Pak could quickly build a similar assessment module and bundle it with their existing MES/SCADA systems, neutralizing Findability's speed advantage. The company's SoftBank backing suggests financial staying power, but the real test will be whether plant operators actually act on the results, regardless of the assessment findings. That remains to be seen.

Read More

B2B Marketing UnBoxed 2026: Date, Speakers, and What to Expect from Bengaluru's AI-Focused Conference

Mavens has scheduled the second edition of its B2B Marketing UnBoxed conference for July 24, 2026, in Bengaluru. The one-day event, themed "Marketing, Disrupted: Own It or Be Outpaced By It," aims to tackle how AI and shifting go-to-market strategies are reshaping B2B marketing leadership.

Confirmed keynote speakers include Parminder Singh, CEO of Reliance Enterprise Intelligences, and Aneesh Reddy, co-founder and managing director of Capillary Technologies. Both are prominent figures in India's tech and marketing landscape. Singh previously served as CMO of Twitter for India and APAC, while Reddy led Capillary through multiple funding rounds totaling roughly $200 million.

The 2025 edition drew over 400 marketing professionals, including more than 250 CMOs and senior leaders, according to Mavens. The organiser has not provided independent verification of those attendance numbers, but the speaker lineup suggests credible draw power for a first-year event.

The 2026 edition has a notable roster of partners: George P. Johnson, CIO Association, Zoho, LinkedIn, Adroit, NeonTrumpet, Wizikey, Xoxoday, Enki Studios, Wozku, and Bmax. This broad sponsorship base — spanning AV, media, gift vouchers, and SaaS platforms — is typical for an event still establishing its identity, though it also risks a fragmented attendee experience if not tightly coordinated.

Context: Where This Fits in India's B2B Event Landscape

India's B2B SaaS sector has grown over 30% year on year, creating demand for events that offer concrete GTM strategies rather than generic MarTech pitches. B2B Marketing UnBoxed enters a field with established players such as the pan-Asia MarTech Summit, which draws 400–600 delegates and has run an AI track since 2024, and the more academic B2B Marketing Leaders' Forum hosted by industry bodies. Digital Marketing Unplugged targets tactical execution while avoiding strategic C-suite framing.

The event's "own it or be outpaced" positioning directly taps frustration among Indian CMOs who, according to a 2025 McKinsey survey, rank AI as a top-three priority. But the conference will need to deliver specific case studies and actionable insights to differentiate from these alternatives and justify the urgency of its theme.

What We Still Don't Know

Mavens has not disclosed registration fees, ticket pricing, or the full agenda. While two high-profile keynotes are confirmed, the session schedule, panel topics, and list of all speakers remain unspecified. The event's website — https://www.b2bmarketingunboxed.com/# — offers no further details as of the July 8 announcement.

Without a detailed program, potential attendees cannot assess whether the content leans toward visionary talks or practical, deployment-based learning. The organisers have not indicated if sessions will focus on AI in marketing execution, GTM restructuring, customer experience case studies, or all three.

Analysis

B2B Marketing UnBoxed 2026 has the right ingredients — timely theme, credible speakers, and a growing Indian SaaS audience — but faces a credibility gap. Mavens itself has minimal public track record; its 2025 inaugural edition was its only prior event. The heavy reliance on 13 partners suggests cost-sharing rather than organic scale. If the agenda remains mostly generic platform pitches dressed as vision talks, the conference risks being seen as a sponsored meetup rather than a substantive industry forum.

A key test will be whether Parminder Singh and Aneesh Reddy share specific, data-backed results from their own AI or GTM transformations, rather than leadership platitudes. For Indian B2B marketers looking to move beyond vendor pitches, the value of this event will depend on how much actual deployment knowledge attendees take home — not just inspiration.

Read More

WZATCO launches Legend GT and Blaze Max projectors on Amazon India Prime Day, starting at ₹24,990

WZATCO, a relatively new name in the Indian projector market, launched two smart projectors on July 8, 2026: the Legend GT and the Blaze Max. Both are available through the WZATCO website and Amazon India during the Prime Day sale. The company is positioning them as affordable options for home entertainment enthusiasts, promising official Google TV, Wi-Fi 6, and auto-focus at competitive prices.

What you get for the price

The Blaze Max is priced at ₹24,990, while the flagship Legend GT costs ₹30,990. Both are introductory offers — the exact duration of the Prime Day pricing isn't specified, but it's described as a limited-time deal. WZATCO says the bundle includes accessories worth ₹3,998, though the company has not detailed what those accessories are. Warranty details and service center locations were not disclosed either, which is a potential concern given WZATCO's limited brand presence in India.

Common features across both models

• Official Google TV with Google Assistant and Chromecast built-in
• Wi-Fi 6 and Bluetooth 5.0
• Google Quick Setup, auto focus, and automatic keystone correction

Legend GT specifics: high brightness, large screen

The Legend GT is the more powerful of the two, claiming 2500 ANSI lumens and the ability to cast an image up to 250 inches. It includes 20W dual stereo speakers, intelligent auto focus, automatic keystone correction, auto screen fit, and what WZATCO calls a "Premium Sealed Optical Engine" for dust protection.

That brightness figure — 2500 ANSI lumens at ₹30,990 — stands out. For comparison, Xiaomi's Mi Smart Projector 2 Pro offers roughly 800-1200 ANSI lumens at a similar or higher price point, and Epson's EF-100 laser projector reaches 2000+ lumens but costs ₹60,000 or more. WZATCO either found a way to offer unusual value, or the measurement may be non-standard. Without independent benchmarks, it's hard to verify.

Blaze Max specifics: convenience-focused

The Blaze Max includes dedicated shortcut buttons for Netflix, YouTube, Prime Video, and Disney+, along with multiple picture and sound modes. Connectivity options include HDMI, USB, Bluetooth, and a 3.5mm audio output. It also has a removable side cover for dust cleaning — a small but practical touch for the Indian environment. The Blaze Max's brightness is not specified in the brief, so it's likely lower than the Legend GT.

How it compares to the competition

WZATCO is entering a market dominated by established brands like Xiaomi, BenQ, and Epson. The Xiaomi Mi Smart Projector 2 series offers similar smart features and a trusted service network, but generally at lower brightness. BenQ's GV30 and GS50 prioritize color accuracy and brand reliability, but their 300-500 ANSI lumens are far lower and prices are higher. Epson's laser projectors deliver high brightness but cost two to three times more and lack Google TV. That puts WZATCO in an interesting spot: it's undercutting the premium options while promising specs that beat mid-range ones. But the lack of a proven track record and service infrastructure is a real drawback.

What we still don't know

Several details were left out of the announcement:

• Processor model, RAM, and storage (key for Google TV performance)
• Exact brightness of the Blaze Max
• Lamp life and warranty terms
• Service center locations in India
• What's actually in the ₹3,998 accessory bundle

These are important bits of information for anyone considering a purchase, especially from a new brand. WZATCO's founder and CEO, Komaldeep Sodhi, said in the announcement that the projectors are "built for the Indian consumer, combining premium features with affordability." That's a marketing statement — the specs do look compelling on paper, but real-world performance and support remain unknown.

Analysis

The Legend GT's 2500 ANSI lumens claim at ₹30,990 is the headline figure, and it deserves skepticism. High-lumen projectors in this price range usually come from brands with little reputation, and the measurement can be misleading (peak versus average, or different testing standards). It's entirely possible the projector will look dimmer in practice than the number suggests, especially with ambient light.

Beyond brightness, the bigger risk is software support. Google TV on third-party devices often stops receiving updates after two or three years, which could leave buyers stuck with an unpatched OS and broken app compatibility. WZATCO hasn't mentioned any update commitment.

Amazon Prime Day is a powerful sales channel, but it also means returns and refunds fall on Amazon's policies, not a local service center. For early adopters willing to take a chance, the price is tempting. For anyone who wants peace of mind, waiting for professional reviews and checking warranty terms before buying is the safer move.

Read More