Google’s Alex Okosi on what’s holding back Africa’s AI startups
On Thursday, Google graduated 15 startups from eight African countries through its Google for Startups Accelerator Africa programme in Nairobi. Most of these startups are building artificial intelligence (AI) into core products in payments, transport, agriculture, healthcare and enterprise software. Google said 60% of the cohort is already profitable, generating an average of $60,000 in monthly revenue. This year’s cohort arrives amid growing debate over whether Africa can turn AI adoption into sustainable, venture-scale businesses. The selected startups offer a snapshot of that transition. Founders are moving beyond experimentation and using the technology to solve operational challenges and build products for local markets. Yet the infrastructure and capital needed to scale those businesses remain in short supply. In an interview with TechCabal, Alex Okosi, Google’s managing director for Africa, said African startups have already embraced AI, but argued that investment has not kept pace. While founders are building AI-powered products and services, the continent still faces gaps in cloud infrastructure, data centre capacity and funding. Those constraints, he said, risk limiting Africa’s ability to capture the economic value created by the technology. AI could add as much as $1.5 trillion to Africa’s economy by 2035, equivalent to roughly 40% of the continent’s current GDP, if governments and private sector players move fast enough to deploy it at scale, according to projections from the African Development Bank. The technology, the bank estimates, could generate hundreds of thousands of jobs while significantly lifting labour productivity across key sectors. The tension between growing AI adoption and limited investment formed the backdrop to this year’s accelerator programme, which featured startups from Kenya, Nigeria, South Africa, Uganda, Tanzania, Senegal, Côte d’Ivoire and Angola. This interview has been edited for clarity and length. Has Africa’s AI moment arrived, or are we still early? If you look at this cohort for Google for Startups Accelerator Africa, many of the companies are AI-first or AI-native because they have built AI into their products to solve real challenges across the continent. That’s the opportunity AI presents, and it’s what excites me about both the startup and fintech ecosystems. Take Mastery Hive, for example. The company is using machine learning to detect fraud across fragmented networks. We also have companies like Loop in South Africa using AI to optimise a complex transit network and manage worker payments. Those are clear examples of startups already embracing AI and putting it to work. That said, the continent still faces challenges. There is a lot of talent and engineering ingenuity in Africa, but infrastructure remains a constraint. Capital is also a major gap. Until investors from the Global North see Africa as a place where AI is being used to solve meaningful problems and deploy more capital here, that challenge will remain. Africa offers significant opportunities to scale and create value. However, it requires investors to view the continent as an AI opportunity rather than a market still waiting to adopt the technology. It’s a twofold story. African companies are already adopting AI and building solutions with it. But from a funding perspective, the level of investment is still not where it needs to be to fully capture the opportunity. What’s the biggest bottleneck to building AI at scale in Africa? Infrastructure remains one of the biggest bottlenecks. To build AI at scale, you need reliable connectivity, and that requires investment in foundational infrastructure such as subsea cables. That’s why projects like Equiano and Umoja are important. They help create the capacity needed for the digital economy to grow. Talent is another critical area. We’ve trained about eight million people in digital skills, giving them the foundations they need to participate in and benefit from the digital economy. We also need to support small and medium enterprises (SMEs). That’s an area we’ve focused on, helping around 35,000 small and medium-sized businesses grow and scale. Those businesses are a key part of the ecosystem. When it comes to AI specifically, compute power is essential. You need cloud infrastructure that developers can build on. That’s one reason we’re investing in our cloud region in South Africa and working to expand cloud adoption. The challenge is that Africa still accounts for only about 1% of global data centre capacity. As a result, many builders have to move data outside the continent for processing and then bring it back to deploy solutions. There’s a clear opportunity to increase local capacity. That will require collaboration between governments, technology companies and other players across the ecosystem. More investment in connectivity, cloud infrastructure and compute capacity will be necessary if we want to scale AI across the continent. Google is playing its part, but this is something that requires collective action from everyone involved in the ecosystem. Must startups have AI at the core of their products to join the programme? No, being an AI company is not a requirement to join the accelerator. That said, AI is an important area for us because we believe it can help startups build solutions faster, scale faster, optimise their operations and reach markets more effectively. We’re encouraging founders to understand how AI can be integrated into their businesses, whether through their workflows, products or business models. That’s why AI training is a key part of the programme. We want companies to understand how they can use the technology to accelerate growth. At the same time, I think we need to demystify AI. It’s not some mystical technology. It’s a tool that helps organisations process information more quickly, analyse large amounts of data and identify opportunities more effectively. For us, technology will play a major role in solving many of the continent’s challenges, and AI is one of the tools that can help make that happen. That’s why we’re continuing to invest in AI skills development. In 2024, we announced a $5.8 million grant programme across Kenya, Nigeria and South Africa to help civil servants and nonprofit leaders build AI capabilities. Through Google.org, we’re also focused on AI skilling across Africa, and
Read More
What we know so far about Samsung Galaxy Glasses
Table of contents When and where Why this launch matters more than usual Design and specs of the Samsung Galaxy Glasses At a glance What’s next: the display model Should you buy the Samsung Galaxy Glasses? Samsung is getting ready to launch its first pair of smart glasses. Inside the company, the device is codenamed Jinju. In public materials, Samsung and Google currently refer to it as Intelligent Eyewear. You will probably know it by a simpler name once it ships, and most reports point to Samsung Galaxy Glasses. This is a brand-new product category for Samsung. The company built these glasses with Gentle Monster and Warby Parker, two eyewear brands you have likely already seen in stores. Here is everything we know about the Samsung Galaxy Glasses so far, including the price, release date, and specs. When and where Samsung is expected to unveil the glasses at Galaxy Unpacked on July 22, 2026, in London. This comes from a report by Seoul Economic Daily, a South Korean outlet with a strong track record on past Samsung announcements. Korea Economic Daily TV picked up the same date. Samsung has not confirmed this date yet. The company usually sends out event invites two to three weeks ahead of Unpacked, so an official confirmation should land sometime in early July. The same event is expected to introduce the Galaxy Z Fold 8, Galaxy Z Flip 8, Galaxy Z Fold 8 Wide, and Galaxy Watch 9 series. That puts five new products on one stage, making it one of Samsung’s biggest hardware days of the year. Even if Samsung shows the glasses at Unpacked, you might not be able to buy them right away. Samsung’s own newsroom says the first collections are scheduled to launch this fall in select markets. Some reports expect a tease in July followed by a proper launch later, similar to how Samsung handled the Galaxy XR headset and Galaxy Ring before they reached stores. Samsung has not announced pricing or availability for the UK, EU, or South Africa yet. Pound and euro prices you might come across online right now are simple currency conversions of the rumored US price, not confirmed regional pricing. Why this launch matters more than usual This launch carries more weight than a typical Samsung product reveal, for a few reasons. This is Samsung’s second Android XR device, after the Galaxy XR headset, which launched on October 21, 2025, at $1,799. Meta already controls most of this market. According to Counterpoint Research, Meta’s share of global smart glasses shipments rose to 82% in the second half of 2025, up from 73% in the first half of the year. AI-powered smart glasses made up 88% of all shipments during that period, and North America accounted for 37% of the market. Samsung is stepping into a space Meta already owns. Samsung also beat Apple to market. Bloomberg’s Mark Gurman reported that Apple’s first smart glasses, known internally as N50, have slipped from a late 2026 introduction to a late 2027 launch. Apple is reportedly working through challenges with its visual AI and Siri, and its glasses are expected to cost between $200 and $500 once they ship. That gives Samsung and Google a big head start. Image source: Samsung Mobile Press Design and specs of the Samsung Galaxy Glasses The first model skips a screen completely. There is no display and no AR overlay. Instead, the glasses lean on audio and a built-in camera that feeds visual information to Google Gemini. Leaked renders from Android Headlines and tipster OnLeaks show a pair of glasses that appear to be regular eyewear. They weigh around 50 grams, with thin temples and a camera built into the frame instead of sitting on top of it. Samsung confirmed its partnership with Gentle Monster and Warby Parker for the glasses. Gentle Monster brings a bold, sharp look, while Warby Parker goes for a more classic style. Both brands showed off frames at Google I/O on May 19 and 20, 2026, and both companies say more styles and prescription lenses are coming later. Here is what reports point to for the hardware, though none of this is officially confirmed by Samsung yet: A 12-megapixel Sony IMX681 camera with autofocus, placed at eye level A Qualcomm Snapdragon AR1 chip, possibly the newer AR1+ Gen 1 version Stereo speakers and multiple microphones for calls and voice commands Bluetooth 5.3 and Wi-Fi, with no built-in cellular connection Photochromic lenses that darken automatically in sunlight A touchpad on the right temple, plus an LED that lights up while the camera is recording On software, the glasses run Android XR with Gemini built in. Gemini can translate signs and conversations as they happen, summarize your notifications, manage your calendar, play music, and take photos when your hands are full. The glasses pair with both Android phones and iPhones, and you can control some functions from a Galaxy Watch. One spec that is still up in the air is battery size. Early leaks pointed to 155mAh. A more recent report from SamMobile, based on a certification filing, found a battery part numbered EB-BO200CAY rated at 245mAh, and Android Authority backed up the larger figure. The most likely explanation is that Samsung is building two different glasses models under the hood, with the smaller battery going to the basic version and the larger one going to a future model with a screen. For this first pair, the 245mAh figure is the better-sourced one right now, though you should treat it as unconfirmed until Samsung says otherwise. Outside estimates put battery life at around 6 to 8 hours of regular use, but that number has not come from Samsung either. At a glance What’s next: the display model A second pair of Galaxy Glasses is already in the works, and this one adds a screen. Reports point to a micro-LED display built into the lenses for true AR overlays, with a launch sometime in 2027 and a price between
Read More
Craydel co-founder Manish Sardana quit a high-flying job to start from zero
The offices of Craydel, a Pan-Africa edtech connecting African students to global universities, occupy a glass-partitioned floor at The Pavilion on Lower Kabete Road, away from Nairobi’s perpetual traffic and construction noise. Through the transparent walls, almost nothing is hidden. Student counsellors are fielding anxious calls from parents and students, while product managers huddle over laptops. Manish Sardana’s office sits in the middle of it all, deliberately so. On one side is the operations team; on the other, the engineers building the artificial intelligence (AI) engine that powers Craydel’s study abroad matchmaking tool. There is no imposing corner office separating the co-founder and CEO from the very people helping him build the company. He asks whether I would like tea. He orders coffee for himself. Sardana has the calm confidence of someone comfortable with uncertainty. He says he has spent his life restless, suspicious of comfort, and constantly searching for purpose. Raised in a modest household in India, he abandoned a place at the prestigious Delhi School of Economics before, over a decade later, walking away from a high-flying career at WPP Scangroup, a marketing and communications company, to build Craydel from scratch. Five years on, he says he has no regrets. He imagines himself still building, still searching for the next problem to solve, until his deathbed. That restlessness has not come cheaply. His family, he acknowledges, has carried part of that burden. “I’ve been married for 18 years now,” he says with a laugh. “So I must have done something right.” When pressed on what success ultimately looks like, he says it is whether his children, his parents, and the people closest to him feel proud of the life he chose to build. This interview has been edited for length and clarity. You grew up in India, built businesses across continents, and eventually chose Africa. What part of yourself were you looking for that you couldn’t find elsewhere? A few things. First, all three co-founders were based in Kenya, so that anchored us here. But personally, I had the option to go back to India and start something there. I chose not to because this continent has been incredibly generous to me. I had sold my company in India and was at a loose end when I got the opportunity to come to Kenya. I built a successful career here, gained a lot, and felt I needed to give back. That was important to me because I see so many expats come, work a few years, make money, and leave. Very few actually stay and contribute to building something lasting. For me, that commitment was real—I invested almost everything I earned and saved here into my venture in Kenya. That was critical. Second, I’m genuinely happy here. Kenyans are warm, and the continent has massive human potential. Yet the number of people solving problems here is very small. In India, there are so many entrepreneurs building so many things—they didn’t need another Manish to start something there. But here, especially in higher education and study abroad, not much was happening. Nobody was disrupting the market; everyone was maintaining the status quo. So we saw an opportunity to do something interesting. I also built startups during my time with ScanGroup on the continent, and I’d turned those ventures into successes. That gave me the confidence to build something here and succeed. Manish (centre), Shayne Aman Premji, co-founder and CFO, and John Nguru, co-founder and CTO. Image source: Craydel Looking back at your twenties, what kind of man were you when nobody was watching? And what parts of that younger man are still alive today? I’m the same person whether someone is watching or not. That doesn’t change how I behave. I’m known for being authentic—for better or worse. In my twenties, I displayed traits like incredible risk-taking. I quit a top college I’d gotten into, in a fiercely competitive environment, and walked away from economics. I had a massive risk appetite then, and I still have it today. I love a challenge. I push myself into corners where I’m really tested. When I’m comfortable, it irritates me; I get bored and seek out new challenges. So, risk-taking, seeking challenges, always looking to build something that creates value—those were the traits I displayed in my twenties, and there’s plenty of evidence of them. I still have them today. When people introduce you, they list your achievements. What do the people who know you best say about you? The people who know me best would say a few things. First, Manish is never easily satisfied—no matter what he’s achieved or received, he wants more; he doesn’t stop. Second, they’d say I have a lot of grit and courage; I’m unshakable. No matter what life throws at me, I stay resilient. Third, they’d say I’m not someone seeking a “chill life.” That’s not me. I don’t seek a life of just joy and ease. And finally, they’d always tell you that Manish needs a very strong purpose to feel satisfied. Without purpose, I feel shallow. Most founders tell a story of opportunity. Yours often sounds like a story of conviction. What is the most expensive belief you’ve ever held onto? The belief that “good is not good enough.” That’s cost me a lot. You achieve something, you feel good, but I never feel it’s enough. So I keep pushing harder, and sometimes it comes at a personal cost; my family has to bear with me. For context, when I quit my job at Scan Group to start Craydel, I was at the peak of my career. I had a great reputation, I was making good money, in a strong position, growing fast. I’d worked incredibly hard to get there. Then I gave it all up to start from zero again. That’s an example of a belief that costs me a lot of money, hardships, and relationships. It just never feels enough. Do you regret it? No, not at all. But I’ve become
Read MoreStarlink grabs headlines, but Safaricom keeps winning broadband users
Safaricom, Kenya’s largest telecoms company, added more broadband subscribers in the first quarter than Starlink has gained in Kenya since launch, underscoring the gap between the attention surrounding satellite internet and its current scale in the market. The telecom operator gained 83,107 fixed internet subscribers in the quarter ended March, raising its customer base to 941,501, according to the latest data from the Communications Authority of Kenya (CA). That compares with Starlink’s total subscriber base of 24,999 customers, which grew by just 2,717 during the period. The figures suggest that while Starlink has captured the attention of regulators, policymakers and rivals, Kenya’s broadband market remains firmly in the hands of operators that have spent years building fibre infrastructure. The data also indicates that incumbents are responding to the satellite threat with faster speeds and revised pricing, intensifying competition in a market where fibre still accounts for the majority of connections. In April, Safaricom doubled speeds on several home fibre packages without raising prices, while rivals including Zuku and Jamii Telecommunications have also revised broadband plans as operators battle for market share. Safaricom’s fixed internet market share rose to 35.4% from 34.9% in the previous quarter, extending its lead over Jamii Telecommunications (JTL), the operator behind the Faiba brand. Jamii added 23,120 subscribers, bringing its customer base to 517,270, though its market share slipped to 19.5% from 20.1%. Several smaller fibre providers also outpaced Starlink’s growth. Vilcom Network added 26,569 subscribers during the quarter, while Ahadi Wireless gained 23,363 customers. The numbers come as Kenya’s fibre providers ramp up competition to defend market share against Starlink, whose entry into the country in 2023 sparked fears that satellite internet could upend the economics of fixed broadband. Yet subscriber growth suggests fibre remains the preferred option for most households and businesses where coverage exists. Safaricom alone added more than three times the number of customers Starlink serves nationwide. However, not all broadband providers benefited from the surge in subscriptions. Poa! Internet, which targets lower-income neighbourhoods with affordable home internet packages, lost 6,788 subscribers during the quarter. Its customer base fell to 256,517, while market share declined from 10.7% to 9.7%. The divergence suggests scale is becoming more important in Kenya’s broadband market. Larger operators are using network reach, bundled services and speed upgrades to attract customers, while smaller providers face pressure from both fibre rivals and satellite entrants. The numbers point to a broadband market where scale still favours fibre operators. Starlink continues to expand in underserved areas, but the industry’s largest players are adding customers at a pace that satellite internet has yet to match.
Read MoreEvery product Google has announced so far in 2026
Table of contents Gemini models Gemini app Google Search Shopping Google Workspace YouTube Android Android XR and Intelligent Eyewear Developer tools Sub Heading 2 Google’s first half of 2026 has been defined by one word: agents. From new Gemini models and an upgraded Search experience to AI-powered shopping tools, Google has spent the first half of the year weaving artificial intelligence into nearly every product it makes. The message from Google’s annual I/O developer conference and other announcements is clear: the company wants Gemini to become the connective tissue across its ecosystem, helping users search, shop, work, create, and complete tasks on their behalf. If you missed the barrage of announcements, here’s everything Google has unveiled in 2026 so far. 1. Gemini models: Google’s biggest announcements of 2026 Google’s AI ambitions took centre stage this year, with the company unveiling new Gemini models built around reasoning, speed, and action. I. Gemini 3.5 Flash Gemini 3.5 Flash is Google’s new default AI model across the Gemini app and AI-powered Search experiences. According to Google, Gemini 3.5 Flash combines frontier-level intelligence with the ability to take action across complex workflows. The company says the model outperforms Gemini 3.1 Pro on coding, multimodal and agentic benchmarks while operating at a significantly lower cost. Google also claims it generates outputs four times faster than comparable frontier models. For users, this could mean faster responses and improved performance on multi-step tasks. The company announced the model on May 19 at its I/O 2026 developer conference and was rolled out the same day. II. Gemini 3.5 Pro Google also previewed Gemini 3.5 Pro, which it positions as the most capable model in the Gemini 3.5 family. While Gemini 3.5 Flash is optimised for speed and everyday use, Gemini 3.5 Pro is designed for advanced reasoning, complex coding, research-intensive tasks and sophisticated agentic workflows. At Google I/O 2026, CEO Sundar Pichai said the model was expected to become available in June 2026. III. Gemini Omni Gemini Omni represents Google’s next step in multimodal AI. Unlike traditional AI models that primarily process text, Gemini Omni can accept images, audio, video and text as input. Google says the model can generate videos grounded in real-world understanding and edit them through natural-language conversations. The company describes Gemini Omni as a system that can “create anything from any input,” starting with video. The first model in the Gemini Omni family, Gemini Omni Flash, is being rolled out through Google’s video creation ecosystem. Google says the model powers video generation and editing experiences in the Gemini app and Google Flow, with the broader goal of making high-quality video creation accessible to users without professional editing skills. 2. Gemini app: From chatbot to personal agent The Gemini app is evolving beyond a simple AI assistant. Google says the app is becoming more agentic, with new capabilities that can use context, learn user preferences over time, and help complete multi-step tasks across Google services. At Google I/O 2026, the company announced new agents and automation features designed to provide more personalised assistance, organise information, and proactively help users accomplish everyday tasks. Gemini Spark Google introduced Gemini Spark, a personal agent designed to complete tasks on users’ behalf. It is currently rolling out to AI Ultra subscribers in the US, with a broader rollout planned for later in the year. Spark can: Work across Google services Continue tasks in the background Handle recurring workflows Personalise assistance based on user context. Daily Brief Daily Brief is Gemini’s personalised morning digest. Drawing on information from Gmail, Google Calendar, tasks, and connected Gemini context, it organises and prioritises the day’s most important items. Beyond summarising information, Daily Brief suggests next steps and actions, helping users stay on top of their schedule and responsibilities. Over time, it can adapt to user preferences to deliver more relevant briefings. Gemini Live upgrades Gemini Live received major updates designed to make interactions more fluid and conversational. Users can switch seamlessly between typing and speaking without losing context, while a redesigned voice experience enables more natural conversations. Google also announced support for regional dialects and deeper integration of Gemini Live across the Gemini app. A redesigned Gemini experience Google introduced Neural Expressive, a new design language for the Gemini app. The refresh brings: Fluid animations Updated typography Improved haptic feedback A more visual interface across Android, iOS, and the web 3. Google Search: The biggest upgrade in years Google Search is evolving from a list of links into an AI-powered assistant. With new capabilities announced at Google I/O 2026, Google Search can answer complex questions, reason across multiple sources, and help users complete tasks directly from the search experience. Search powered by Gemini 3.5 Flash Google has integrated Gemini 3.5 Flash into its AI-powered Search experiences, including AI Mode and AI Overviews. This upgrade enables more complex queries and more comprehensive, context-aware responses across Search. Interactive answers Search will increasingly present AI-generated responses alongside traditional results, with more interactive and visually rich formats. Depending on the query, users may see dynamic layouts, visual elements such as charts or images, and explanations tailored to the context of their question. Long-running tasks Google is introducing agent-like capabilities in Search that support multi-step and ongoing tasks, allowing users to continue and track certain workflows over time. Search is increasingly designed to help users complete tasks as well as find information. 4. Shopping: Google’s buying assistant Shopping is also receiving the AI treatment. Google wants to help users move from discovery to purchase with minimal friction. Universal Cart Universal Cart is one of the biggest new Shopping announcements. It is a Gemini-powered shopping cart and agentic hub available across Google Search, Gemini, YouTube, and Gmail (with integrations rolling out progressively). Once you add a product, Google can proactively: Monitor price drops Track price history insights Surface deals automatically Alert users about stock availability Recommend alternative or similar products It is also designed to better understand shopping context, including: Relevant merchant offers and promotions Payment-related incentives within
Read MoreFintech Yoco thinks South Africa’s small businesses need fewer apps
Yoco, a South African fintech company, built its business by helping small merchants accept card payments. Now it wants to help them run their entire businesses. At its Yoco Next 2026 event in Johannesburg on Tuesday, the company unveiled more than 20 new products and features, including AI-powered business tools, loyalty programmes, savings products, accounting integrations, industry-specific software and a R250 million ($15.2 million) annual reduction in transaction fees for merchants. Carl Wazen, Yoco’s co-founder and chief business officer, said the announcements reflect a broader shift in strategy. The startup now wants to provide the software that small businesses use to manage their operations. From inventory and accounting to bookings, customer loyalty, cash flow and reporting, many business owners rely on multiple disconnected tools. Yoco’s goal is to bring those functions into a single platform. “Yoco started by giving independent businesses access to payments,” said Wazen. “Today, we are giving them the tools that used to belong only to big business, at a price built for small business.” Founded in 2015, the fintech serves more than 200,000 merchants across South Africa and offers payment devices, software, business financing and commerce tools. But executives say payments are becoming just one part of the business. “We are no longer just a payments company,” chief executive officer (CEO) Carsten Höltkemeyer told TechCabal. “We are a company that helps business owners reduce administrative burdens and simplify everyday operations through modern technology and innovation.” One of the innovations unveiled on Tuesday is Yoco AI, an artificial intelligence assistant expected to launch in the third quarter of 2026. Developed after the company acquired Dyner.ai, an AI-native operating system built to help small and medium-sized enterprises (SMEs), in May, the tool will analyse transaction data, customer behaviour and business performance to help merchants make decisions about their businesses. “If you can use data to better understand your revenue streams, customer behaviour and business performance, you’re in a stronger position to succeed,” Höltkemeyer said. “We want to give merchants access to the kinds of insights and capabilities that large corporations have traditionally enjoyed.” The company’s CEO also unveiled the Yoco Connect, a hub that links accounting software, e-commerce platforms and inventory systems. It has launched dedicated software for restaurants, retailers, salons, and wellness businesses, each designed around the operational needs of its sector. For Kelly Gibberd, founder of Cape Town-based fashion retailer Me&B, managing a growing business involves far more than processing payments. Gibberd’s company employs 55 people, supports 10 local factories, operates physical stores and runs an e-commerce platform. As a manufacturer and retailer, Gibberd said rising costs and increasing competition have made operations more complex. “Manufacturing in South Africa comes with limitations; costs remain a huge challenge, and customers compare us to global fast-fashion giants like Shein and Temu,” she said. Tools that simplify inventory management, customer engagement, reporting, and cash flow management could help businesses spend less time on administration and more time serving customers, Gibberd added. Alongside the new software products, Yoco said it has reduced transaction fees by up to 40% across parts of its network. “It’s an investment in our customer base,” Höltkemeyer said. “Reducing our fees is an investment in small businesses, but we also believe the new products and services we are launching will create a much bigger opportunity for both Yoco and our merchants.”
Read MoreGemini 3.5 Flash vs Gemini 3.1 Pro: What’s the difference?
In May, Google launched Gemini 3.5 Flash, the first model in its new 3.5 family of AI models. The model was built to be faster and more capable in handling tasks that require an AI to take actions. The company said it performs well at coding, using tools, reasoning across different types of information, and completing multi-step tasks. Google’s AI lineup can feel confusing from the outside. There are different version numbers, Flash models, and Pro models, but there is a logic to it. Because not every user needs the same thing from an AI model, Google builds different versions for different purposes. The Flash models are built for speed and efficiency, while the Pro models are built for deeper reasoning and more demanding analytical work. If you use Gemini frequently, your next question will probably be: if Gemini 3.1 Pro already exists, what is Gemini 3.5 Flash supposed to do differently? This guide breaks down the differences and will help you figure out which model makes sense for how you use AI. What is Gemini 3.5 Flash? Gemini 3.5 Flash, like all Flash models, was designed for speed and efficiency, and has a more recent knowledge cutoff of January 2025. This means it is better informed about recent events when it’s answering from its training data. What is Gemini 3.1 Pro? Gemini 3.1 Pro is Google’s previous flagship, released in February 2026. It was built with deep reasoning at its core and is the kind of model used when a task requires multi-layered thinking rather than fast responses. Its clearest strength over 3.5 Flash is its ability to process large volumes of information and documents while maintaining context across lengthy conversations. So, what’s the difference? According to benchmarks published by Google alongside the release of Gemini 3.5 Flash, the newer model outperforms Gemini 3.1 Pro in several practical tasks. However, the benchmarks show that Gemini 3.1 Pro still holds advantages in some areas. Coding and software development: For developers and people who use AI to write code, Google’s benchmarks show that 3.5 Flash has stronger performance across multiple coding evaluations, including software engineering tasks, code generation, and debugging challenges. In tests that put AI models through real coding tasks in a terminal environment, Flash scored 76.2% compared to Pro’s 70.3%. Agentic tasks and tool use: Agentic tasks are tasks where the AI needs to do more than answer a question, like conduct a search or complete several actions before arriving at a final answer. Google’s testing shows Gemini 3.5 Flash performs noticeably better in these situations, suggesting it is better suited for AI assistants and automated workflows. On tests measuring multi-step, tool-assisted performance, Flash scored 83.6% compared to Pro’s 78.2%. Research, analysis, and professional tasks: Google also found improvements in specialised tasks involving financial analysis and decision-making. While some users may not notice the difference in everyday conversations, professionals using AI for research or financial modelling may benefit from Flash’s stronger performance in these areas. Flash scored 57.9%, while Pro scored 43% in the benchmark test. Long-document performance: This is one of Gemini 3.1 Pro’s strongest remaining advantages. When the task involves finding specific information buried deep inside a very long document, Pro remains more accurate. In testing on documents around 128,000 words long, Pro scored 84.9% while Flash scored 77.3%. If your work regularly involves analysing lengthy reports or research papers, Gemini 3.1 Pro may be the better option. Reasoning: On tasks that test pure reasoning ability, such as complex logic problems and abstract pattern recognition, Gemini 3.1 Pro still holds an edge. In a test designed to challenge the limits of AI reasoning across academic subjects, Pro scored 44.4% to Flash’s 40.2%, while Pro scored 77.1% to Flash’s 72.1% in a test of abstract reasoning puzzles. Which model should you use? The biggest difference between Gemini 3.5 Flash and Gemini 3.1 Pro is what they are optimised for. For everyday users, Gemini 3.5 Flash will likely be the more practical choice, as it is faster and performs well across a wide range of tasks. If your work depends on long-context understanding or solving difficult reasoning problems, Gemini 3.1 Pro remains one of Google’s strongest models. Neither model replaces the other; it just depends on what you need it to do.
Read MoreSamsung Galaxy Z Fold 8 Wide: Release date, price, and specs
Table of contents When and where is Samsung announcing the Galaxy Z Fold 8 Wide? Why this launch matters more than usual Is it called the Wide or just the Galaxy Z Fold 8? Samsung Galaxy Z Fold 8 Wide specs Samsung Galaxy Z Fold 8 Wide: Specs at a glance Samsung Galaxy Z Fold 8 Wide vs. Galaxy Z Fold 8 Ultra: What is the difference? How much will the Samsung Galaxy Z Fold 8 Wide cost? Should you buy the Samsung Galaxy Z Fold 8 Wide? Samsung is about to change how its foldable phones are named, how many book-style foldables it sells at once, and possibly how the rest of the industry thinks about wide foldables. The Samsung Galaxy Z Fold 8 Wide comes as Apple is preparing its first foldable iPhone. Samsung is not waiting around. Here is everything you need to know about the Galaxy Z Fold 8 Wide ahead of its expected launch. When and where is Samsung announcing the Galaxy Z Fold 8 Wide? Samsung is widely expected to announce the Galaxy Z Fold 8 Wide at Galaxy Unpacked on July 22, 2026, in London. The date was first reported by Korea Economic TV and has since been confirmed by SamMobile, Android Authority, Tom’s Guide, and Android Police, among others. As of mid-June 2026, Samsung had not issued an official media advisory, so treat this as a very strong rumor rather than a done deal. The choice of London is significant. Samsung has historically held its summer Unpacked events in Seoul, New York, or San Francisco. London puts Samsung front and center in Europe, one of Apple’s strongest premium markets, just months before Apple is expected to announce its own foldable iPhone. If Samsung follows its usual pattern, pre-orders will open the same day as the announcement, with devices shipping to customers roughly two weeks later, putting availability in early August 2026. Alongside the Wide, Samsung is expected to announce: Galaxy Z Fold 8 Ultra (the traditional tall foldable, successor to the Fold 7) Galaxy Z Flip 8 Galaxy Watch 9 series Galaxy Glasses, Samsung’s first AI smart glasses running Android XR Why this launch matters more than usual Three things make the Galaxy Z Fold 8 Wide more than just another Samsung hardware refresh. First, Apple’s foldable iPhone is coming. Widely referred to as the “iPhone Fold” or “iPhone Ultra,” Apple’s first foldable is expected in September 2026 at a starting price above $2,000. Samsung’s July launch gives the Wide a roughly two-month window in the market before Apple ships a single unit. Apple’s device is also rumored to use a wide, near-4:3 design, which means the two phones will be aimed at the same type of buyer. Second, Samsung is launching two book-style foldables simultaneously for the first time. Previous Unpacked events gave you the Fold or the Flip. This year, you get the Wide, the Ultra, and the Flip 8 all at once. Samsung is splitting its Fold line into two distinct devices: a wide, lighter model for media and multitasking, and a taller, camera-rich model for power users who need zoom and the biggest screen. Third, Samsung is betting heavily on this device. Korean supply chain reports citing ETNews say Samsung initially planned around 1 million units and then increased that by 200,000 to 300,000, putting the Wide’s production at parity with the standard Fold while trimming Flip 8 output. Samsung does not quietly add to production runs for devices it is unsure about. Is it called the Wide or just the Galaxy Z Fold 8? This is where things get confusing, and you deserve a straight answer before you go any further. For most of the pre-launch leak cycle, industry insiders referred to the two 2026 foldables as the “Galaxy Z Fold 8″ (the tall successor to the Fold 7) and the “Galaxy Z Fold 8 Wide” (the new wide model). Late in development, Samsung appears to have flipped the naming. A Bluetooth SIG certification filing confirmed the name “Galaxy Z Fold 8 Ultra” for model SM-F976, which is the tall Fold 7 successor. That means the wide model, SM-F971U, is expected to launch simply as the “Galaxy Z Fold 8.” So the device this article is about will very likely be sold at retail without the word “Wide” in its name at all. This article uses “Galaxy Z Fold 8 Wide” throughout to avoid confusion with the Ultra, but when you walk into a store, the wide model will probably just say “Galaxy Z Fold 8″ on the box. Samsung Galaxy Z Fold 8 Wide specs 1. Display The inner display is a 7.6-inch LTPO OLED panel with a 4:3 aspect ratio. This is the first Samsung foldable to use a 4:3 inner display, making it shorter and wider than any Fold that came before it. Think iPad mini proportions rather than a tall, narrow smartphone screen. Both panels support 1-120Hz adaptive refresh, HDR10+, and a peak brightness of up to 2,600 nits. The cover screen is 5.4 inches with a 4.7:3 aspect ratio. That wider cover screen is a meaningful improvement over older Fold cover displays, which were so narrow they were barely useful for anything beyond checking notifications. 2. Processor and performance The Wide runs on the Snapdragon 8 Elite Gen 5 for Galaxy, the same overclocked chip found in the Galaxy S26 Ultra. This is confirmed across virtually every leak and supported by FCC filings. No Exynos variant has been reported for the 2026 Fold line. RAM options are 12GB or 16GB (LPDDR5X), with storage at 256GB, 512GB, or 1TB (UFS 4.0). There is no microSD slot. Some reports suggest the Wide may top out at 512GB while the 1 TB option is reserved for the Ultra, but this is unsettled. 3. Battery and charging The Wide is expected to carry a roughly 4,800mAh battery, which would make it the largest battery ever put in a Samsung Fold-style device. For comparison, the Ultra
Read MoreWhy global investors keep missing Africa’s biggest climate opportunity
This article is based on a conversation from Voices & Visions, a podcast produced through a partnership between Tutto Passa Agency and TechCabal, which explores the people and ideas shaping Africa’s innovation economy. The world’s biggest climate challenge at the moment may not be raising more money, but deploying the existing capital differently. In Africa, investors have committed $44 billion annually, up nearly 50% from a few years ago, yet founders building technologies to help farmers survive droughts, businesses reduce emissions, and communities adapt to changing weather patterns still struggle to access capital. According to Victor Ndiege, the CEO of Kenya Climate Ventures (KCV), an impact investment manager focused on early-stage climate enterprises, their problem is not a shortage of investors but funds unwilling to finance risk on local terms. The disconnect is exposing one of the weaknesses in Africa’s green transition. As climate adaptation becomes the top policy agenda, local fund managers argue that the financial instruments meant to support it are designed for mature markets and not small local enterprises. “There are many investors who have not been able to deploy capital, not because there are no businesses requiring capital, but because of the terms and structures around deploying that capital,” Ndiege said. “The financial architecture is the most important aspect of investing, not the name of the instrument.” Long game While billions of dollars have been committed globally to climate action, only a tiny share reaches businesses helping communities adapt to changing weather patterns. Climate adaptation—the technologies and services that help economies cope with changing rainfall patterns, water shortages, and rising temperatures—remains one of the least financed segments of climate investing. Many climate businesses require long-term investments before generating stable cash flows. For example, farmers adopting new irrigation technologies or households switching to solar power might not produce venture-scale returns overnight. Their growth relies on patient deployment and steady operational growth. Ndiege argues that this mismatch comes from the assumptions investors bring to Africa. His criticism extends beyond venture capital to the development finance ecosystem. KCV is currently raising a $25 million climate fund after building a revolving investment facility over the past decade. However, several development finance institutions have indicated that the proposed fund is too small for their participation. “$25 million is what we need to grow sustainably,” Ndiege said. “But many investors would rather wait until we are managing $100 million before coming in. By then, we may no longer need them.” This observation exposes a paradox that is alive in most startup ecosystems around the world. Institutions that are meant to finance early-stage startups only back mature fund managers and do not help smaller ones scale. Most ecosystems are built in such a way that capital follows validated successes. Meanwhile, local entrepreneurs continue struggling to secure financing. Local currencies The firm invests for as long as seven years, allowing companies to mature before repayment obligations intensify. Funding is disbursed in stages, matching the pace at which businesses grow. One area where KCV claims it has broken with most of its peers is currency. It deploys local currency financing and not dollars or pounds. “If we chose to invest in dollars or pounds, businesses would spend more money hedging against currency fluctuations,” Ndiege said. “We want entrepreneurs to spend their time growing their businesses rather than responding to risks that we can help them overcome.” For African businesses generating revenues in local currencies, exchange-rate volatility can turn otherwise viable investments into distressed assets. The issue has become important as many African currencies have depreciated against the dollar over the past three years, increasing repayment burdens for companies financed in foreign currency. KCV also rejects the argument that early-stage companies should avoid debt entirely. According to Ndiege, most investors assume that only equity financing is appropriate for young businesses. In 2025, African startups raised a record $1.64 billion to $1.8 billion in debt financing. “The question is not whether debt works,” he said. “The question is how you structure that debt facility to respond to the growth of that startup.” Repayment schedules, interest costs, and deployment timing, he argues, should reflect business life cycles and not banking rules. “If our portfolio companies do not succeed, then we have not succeeded,” Ndiege said. “It should not be an investor-investee relationship. It should be a partnership for growth.” This support should be extended to other operational needs of a young company. Many African startups fail not because the demand is absent but because they lack audited financial statements, governance structures, or investment documentation required by institutional investors. “The challenge is translating what entrepreneurs know into a language investors understand,” Ndiege says. “It is not about their ability to do business. It is about compliance.” KCV’s investment combines capital with technical assistance, governance support and managerial development. The approach is resource-intensive but reflects the realities of African enterprise growth, where many businesses emerge from informal markets before formalising. If climate finance continues to reward only businesses that have already succeeded, Africa may discover that the biggest barrier to climate adaptation was never the availability of money, but the way capital itself was designed. Listen to the full podcast on Spotify.
Read MoreUkiyo’s new app connects South African students to their first jobs
Ukiyo, a South African educational technology company, has launched a mobile platform it says addresses a gap in how young people in the country access opportunities, from bursaries and accommodation to internships and mental health support. The Global Student Support Platform (GSSP) combines scholarships, job listings, mentorship, wellness services, tutoring, and career development tools in a single app. Ukiyo is positioning it as a marketplace for youth development services, free for students to use. The launch comes as South Africa‘s youth unemployment crisis deepens. In the first quarter of 2026, the unemployment rate of youth aged 15 to 24 in South Africa stood at 60.90%, while around 3.9 million young people in the same age bracket are classified as not being in employment, education, or training (NEET). For Nozuko Mzamo, founder of Ukiyo, the issue is a lack of systems that connect ambitious young people to opportunities. “South Africa does not have a shortage of ambitious young people. It has a shortage of integrated pathways into economic participation and systems that connect young people to what they need to succeed,” Mzamo said in a statement. “We built GSSP to support the full journey, from finding a place to study and securing education funding, to building a career and accessing mentorship.” On GSSP, students can explore study options, access mentorship opportunities, connect with accommodation providers, find wellness and mental health support, attend skills development workshops, and search for internships, graduate programmes, and entry-level jobs. Founded in 2017 after operating informally since 2014, Ukiyo was created to tackle youth unemployment by addressing some of its underlying causes, including limited access to information and skills development. Mzamo said the idea took shape after she observed that opportunities discussed in corporate boardrooms rarely reached the students who needed them. Ukiyo built GSSP to close that information gap, particularly for young people outside major urban centres. The company noted that GSSP has attracted more than 4,200 users who have generated over 1,300 click-throughs to bursary and scholarship opportunities and more than 2,100 click-throughs to job listings in its private beta. Users currently discover opportunities through filters and searches, but Mzamo noted that Ukiyo plans to introduce intelligent matching features in future releases. “Whether you’re figuring out what to study and where, looking for bursaries or scholarships to fund your studies, needing academic, psychosocial or career readiness support, hunting for student deals, or searching for your first job, GSSP walks with you through every one of those milestones and equips you with the skills and information you need at each stage,” Mzamo said. Ukiyo works with corporate partners, including higher education institutions, funders, employers and service providers to bring opportunities and support services onto the platform, according to Mzamo. Some of its current partners include Thrive Accommodation, North-West University, The LINK by Airlink, and Emeris. With its launch, GSSP competes with platforms such as LinkedIn, Pnet, and Jobox that help students and graduates discover internships, graduate programmes, and entry-level jobs. However, Mzamo argued that most existing services focus on only one part of the student journey. “That’s exactly the distinction: they address one or two pieces of the puzzle,” she said. “GSSP covers the full cycle, from studies to funding, to skills development, to finding that first job.” Mzamo noted that the platform had already started including global exchange programme opportunities, with plans to expand its scope beyond South Africa. “We’re expanding our research to cover Pan-African and broader international markets, but for now, the opportunities on GSSP are for South African youth, whether they choose to stay, study, or work here or abroad,” she said.
Read More