Current Trends in Startups and Venture Investment as of August 27, 2026: Record $510 Billion in Half a Year, Mega-Rounds for Gatik and Emerald AI, Nvidia's Deal with Poolside, Sale of Hugging Face, Shein's IPO in Hong Kong, and Anthropic's Preparation for the Largest Offering in History.
The venture market is approaching the end of August 2026 in a state that can only be described as paradoxical. On one hand, global investments in startups reached a record $510 billion in the first half of the year—more than in all of 2025. On the other, money is distributed extremely unevenly: in the U.S., artificial intelligence captured 86% of venture dollars in the second quarter, while the Federal Reserve's rate remains at 3.50–3.75%. The result is a market where "dry powder" coexists with stringent selection.
The major shift in recent days is that capital has stopped paying merely for "AI exposure." Investors are acquiring control over the constraints created by the adoption of artificial intelligence: energy for data centers, safety for autonomous models, chip design, licensed content, and physical logistics. Below are the key events and trends shaping the agenda for venture investments on Thursday, August 27, 2026.
- Record semi-annual volume and concentration of capital. $510 billion globally, over $400 billion in the U.S., with mega-rounds and AI driving almost all dynamics.
- Mega-rounds in AI "bottlenecks." Gatik raised $200 million for autonomous freight transport, Emerald AI—$150 million in Series A at a valuation of $1.05 billion, and Alice—$140 million for model safety.
- Strategic capital instead of classic M&A. Nvidia pays Poolside $6 billion for a license and another $1 billion for a stake; labels and Electronic Arts are entering Stability AI.
- Revival of exits. Hugging Face is exploring a sale at a valuation of $13 billion, Shein is going public in Hong Kong, and Anthropic is preparing to file an S-1 by the end of the month.
- Local focus: Russia and CIS. The market is shrinking in volume but growing in deal quality—the median check increased by 23%.
Macro Update: Record Capital Amidst High Rates
According to Crunchbase, global venture investments from January to June 2026 totaled $510 billion, compared to $440 billion for all of 2025. PitchBook-NVCA data shows that American startups attracted over $400 billion in the half-year, with AI accounting for 86% of all venture dollars in the second quarter. Physical AI—robots, autonomous systems, drones—collected more in six months than in the combined total of 2022–2024 ($41.9 billion).
Meanwhile, the Fed maintained the 3.50–3.75% range at its July meeting, with three committee members advocating for a hike. This fundamentally differentiates the current boom from that of 2020–2021: venture funds are deploying record amounts without the support of zero rates. For investors, this signifies "two-tier" dynamics: exceptional companies with access to structural demand for AI are receiving extraordinary valuations, while undifferentiated software is facing tough follow-on rounds.
Deal of the Day: Gatik Raises $200 Million for Autonomous "Middle Mile"
Gatik from Santa Clara closed a $200 million Series D round led by Qatar Investment Authority and Koch Disruptive Technologies, with participation from Millennium Management, ARK Invest, and Intact Private Capital. The company focuses on autonomous freight transport between distribution centers and retail locations—repeating routes instead of the open task of robo-taxis.
Why This Round Matters for Venture Investors
- Over $600 million in contracted revenue and 85,000 fully autonomous deliveries—rare commercial validation for the sector.
- Total capital raised—approximately $500 million; new valuation was not disclosed.
- Qatari sovereign capital partnering with an industrial investor like Koch signals that capital-intensive physical AI is financed with contractual demand in mind.
Emerald AI: Unicorn at Series A and Energy Hub of AI Infrastructure
The most notable price signal of the week is the $150 million Series A of Washington's Emerald AI at a valuation of $1.05 billion. The round was led by Energize Capital and DCVC, with a syndicate reminiscent of a map of industry interests: NVIDIA, Samsung Ventures, Siemens, Aramco Ventures, Salesforce Ventures, GE Vernova, RWE, JERA Ventures, In-Q-Tel, Lowercarbon Capital.
Emerald Conductor allows data centers to flexibly vary energy consumption based on grid conditions without interrupting computations. According to the company, the approach could "unlock" over 100 GW of existing capacity in the American energy system. The company is valued at Series A based on the scale of the constraint it addresses—this is the new logic of pricing in AI infrastructure.
Model Safety and Content: Alice and Stability AI
Alice (formerly ActiveFence) raised $140 million led by Apax Digital with participation from Samsung and SentinelOne, bringing total funding to $280 million. The company collaborates with eight out of ten leading AI labs and is approaching $100 million in annual recurring revenue; valuation ranges from $800 million to nearly $1 billion. The thesis is simple: as models transition from responses to actions within corporate systems, AI safety becomes a distinct category alongside network and identity security.
Stability AI closed a $76 million Series B, where the importance lies not in the amount but in the composition of investors: Universal Music Group, Sony Music Group, Warner Music Group, and Electronic Arts joined alongside AMD Ventures. Rights holders are transforming from plaintiffs to shareholders—financing becomes a component of corporate architecture that mitigates licensing risk.
Strategic Capital: Nvidia Reshapes AI Deal Rules
Nvidia's deal with Poolside—$6 billion for a non-exclusive license on the Model Factory system plus $1 billion investment at a pre-money valuation of $12 billion and transferring over 100 engineers to the Nemotron open models project—sets a new template. Instead of classic acquisitions, corporations utilize licenses, minority stakes, and talent deals. The seller's story is telling: Poolside failed to raise $2 billion in six weeks for a cluster of 40,000 GPUs and lost it. Access to computation has become the primary filter for survival for second-tier models.
Simultaneously, Nvidia is negotiating investments in Perplexity at a $30 billion valuation (with revenue exceeding $750 million) and in Mercor at $20 billion. The same pattern is observable in the day’s deals: Builders FirstSource solely financed a Series A of $25.3 million for startup Digs and signed a five-year commercial contract; Tencent led a Series B of $18 million for Dublin-based W4 Games with a commitment to develop the Godot ecosystem in Asia.
M&A and IPO: Exit Window Expands
- Hugging Face has hired a bank to assess buyer interest at a valuation starting from $13 billion—almost three times higher than the $4.5 billion in Series D funding from 2023. This continues the wave of reevaluation for the "distribution layer" of AI following Stripe's acquisition of OpenRouter for more than $7 billion.
- Shein is conducting its IPO in Hong Kong: up to $1.77 billion at a valuation of approximately $27 billion—down from $100 billion at its peak. The price will be announced on August 31, with trading starting on September 1 after unsuccessful listing attempts in New York and London.
- Anthropic is preparing to file publicly by the end of August with a target valuation of about $2 trillion, with an offering volume comparable to SpaceX's record IPO. The IPO volume in the U.S. since the beginning of the year stands at $160.6 billion, with a historical high of $195.2 billion in 2021.
Physical AI and Asia: From Guangzhou to Seoul
XPeng's robotics division raised over $900 million in its first external round at a valuation exceeding $6 billion, with participation from IDG Capital, Tencent, and Alibaba—the company plans to produce around 1,000 humanoids named IRON per month by the end of the year. In India, Airbound raised $37 million in Series A led by Greenoaks for autonomous flying vehicles, MATTER Motor Works secured $25 million, and wealthtech platform Nexedge raised $20 million. In Korea, Liner closed a Series C of $36.1 million mainly from local institutional investors, building a layer of verifiable AI search for corporations.
Russia and CIS: Fewer Deals, Higher Standards
The Russian venture market is moving in stark contrast to the global trend: according to the Moscow Venture Fund, investments for the first half of 2026 totaled 4.6 billion rubles across 54 deals, but the median check increased by 23%—to 24.6 million rubles. The high key rate has made deposits a rational alternative to long-risk assets, and investors have completely abandoned financing "promising ideas" without revenue. Growth points include corporate funds in medicine and industrial technologies, as well as regional platforms like the Siberian Venture Fair.
What to Watch for Investors on August 27
- Reaction to Nvidia's Report. Results for the second quarter of the 2027 fiscal year were released after trading closed on Wednesday; consensus expected revenue of around $92 billion (+97% year-on-year). Stock dynamics on Thursday will set the tone for valuations across the entire AI infrastructure.
- Kickoff of the Symposium in Jackson Hole. Fed signals regarding rate trajectories directly impact capital costs for late-stage rounds and the IPO pipeline.
- Anthropic's public filing and Shein's price announcement on August 31—two tests of the public market's appetite for AI and for "tired" unicorns, respectively.
Conclusion: They're Not Paying for Models, But for Scarcity
The agenda for August 27, 2026 confirms that the venture market has entered a phase where capital is concentrating around strategic scarcity. Electricity, agent safety, chip design, content rights, and contractual logistics are funded more generously than yet another interface for a substitutable model. For venture funds, this signifies a reevaluation of portfolio construction: the question for startups is no longer "where is the AI here?" but "what scarce resource does the company control, and will its advantage survive the cost reduction of the models themselves?"