With venture capitalists raising larger funds and moving to later stage investments, there are many promising innovations going unfunded - despite the promise of higher returns from being in at the beginning.

The traditional model for innovation was to fund start-ups based on academic research, but a lack of appetite for risk and the lure of appreciative markets for proven products means many basic research projects never make it to the next stage.

"Google is the exception, rather than the rule," says Patrick Ennis, managing director of Arch Venture Partners, which has invested exclusively in seed and early-stage ventures for 19 years.

Investment in seed and start-ups has fallen precipitously as a percentage of total venture-capital investment since the heady days of 1999, according to quarterly figures from National Venture Capital Association.

Starting off in the first-quarter of 1999 with nearly 9% of total venture investment, seed and start-up investment had fallen to 7.2% of the total by the third quarter of 1999 and to 3.61% by year end. Then, after being at 2% to 3% for 18 months, the percentage of total venture-capital investment represented by this initial-stage investing hovered between 1% and 2%, and stood at 1.9% at the end of the first quarter of this year.

Real early-stage investing involves taking the fruits of academic labor and creating the mechanism to commercialize the product, turning it from a concept into an operational business. Basic research grants will cover the formation of a company but after that, outside investment is needed to develop the laboratory results to a point where a product can be produced and to work on licensing deals from the research institutions that typically get cash and equity in return for the license. Funding is also needed to rent office space and hire engineers.

It's a labor-intensive job and larger funds of, say, $500 million can't afford to spend disproportionate amounts of time on small investments - seed and early-stage investment is anything from around $500,000 to about $5 million.

Arch Venture has invested in the earliest capital rounds of more than 110 companies out of its six funds, which have a combined total of $1 billion, commercializing technology that is developed at universities and national laboratories and occasionally at large research companies.

Its initial investment is always at the seed stage and the venture capitalist will support the company through to its initial public offering or other exit, such as a private sale, participating in later rounds with other investors. The investment period can be anything between three and seven years and its commitment isn't altrusitic, says Ennis - we stay in because otherwise we wouldn't be enjoying the increase in the company's value as it grows.

One example is Arch Venture's five-year-long investment in Impinj Inc., a fabless semiconductor company which makes very low power computer chips for use in RFID, or radio frequency identification, technology. Developed at the California Institute of Technology by semi-conductor guru Carver Read and also in Seattle with co-founder Chris Diorio, professor of computer science and engineering at the University of Washington , the company's technology is used for consumer electronics devices, industrial control and PC peripheral devices worldwide.

Patience and risk-taking are rewarded by the superior gains enjoyed by seed and early-stage funds compared with returns on balanced and late-stage funds. Long-term performance returns for initial-stage venture capital at the end of 2004 was 19.9% on a 20-year investment period, 44.7% on a 10-year period and 38.9% on a one-year period, according to the NVCA. Comparable returns for balanced venture capital were 13.3%, 18.2% and 14.7%, and for later stage 13.7%, 15.4% and 10.4%.

Boston-based PureTech Ventures was set up in 2001 specifically to mine academia for gems in the life sciences sector. Describing the firm as a "sort of hybrid between a venture fund and an entrepreneur," founder and Managing General Partner Daphne Zohar says they look at about 1,000 academic projects each year and invest in two or three of them - "there is a backlog of brilliant ideas out there," she adds.

Its investments to date include Cellicon Biotechnologies Inc., which has developed a technology that has the potential to restore the effectiveness of current antibiotics that have become ineffective because of bacteria resistance. With PureTech's investment, Cellicon is in the process of setting up pre-clinical "in vivo" trials.

Zohar is also keen to capitalize on the massive aesthetics market and many of the technologies PureTech is invested in will likely have the potential to be used in this sector in addition to their therapeutic or other medical applications.

PureTech's initial funding in any investment comes from the partners in the firm. As well as Zohar these include a number of industry experts such as co-founder John Zabriskie, the former chief executive of Pharmacia & Upjohn; Ben Shapiro, who was previously executive vice president for Merck & Co.'s (MRK) worldwide licensing; and co-founder Bob Langer, who is professor of chemical and biomedical engineering at MIT.

Zohar says the firm works very closely with top-tier venture capitalists and will syndicate with these firms for follow-on investments. For example, Boston Millennia Partners, IDG Ventures, S.R. One Limited and Novo Ventures joined in a $19 million financing round in Protein Forest in November 2003, a company that makes protein-separation products and that was co-founded by PureTech in 2002 with Israeli physicists Drs. Shmuel Bukshpan, Uri Halavee and Gleb Zilberstein.